Merge branch 'w-okada:master' into master

.github/ISSUE_TEMPLATE/issue.yaml

@@ -117,6 +117,16 @@ body:
     id: issue
     attributes:
       label: Situation
-      description: Developers spend a lot of time developing new features and resolving issues. If you really want to get it solved, please provide as much reproducible information and logs as possible. Provide logs on the terminal and capture the window.
+      description: Developers spend a lot of time developing new features and resolving issues. If you really want to get it solved, please provide as much reproducible information and logs as possible. Provide logs on the terminal and capture the appkication window.
+  - type: textarea
+    id: capture
+    attributes:
+      label: application window capture
+      description: the appkication window.
+  - type: textarea
+    id: logs-on-terminal
+    attributes:
+      label: logs on terminal
+      description: logs on terminal.
     validations:
       required: true

.github/workflows/cla.yml

@@ -0,0 +1,36 @@
+name: "CLA Assistant"
+on:
+  issue_comment:
+    types: [created]
+  pull_request_target:
+    types: [opened, closed, synchronize]
+
+jobs:
+  CLAssistant:
+    runs-on: ubuntu-latest
+    steps:
+      - name: "CLA Assistant"
+        if: (github.event.comment.body == 'recheck' || github.event.comment.body == 'I have read the CLA Document and I hereby sign the CLA') || github.event_name == 'pull_request_target'
+        # Beta Release
+        uses: cla-assistant/github-action@v2.1.3-beta
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          # the below token should have repo scope and must be manually added by you in the repository's secret
+          PERSONAL_ACCESS_TOKEN: ${{ secrets.PERSONAL_ACCESS_TOKEN }}
+        with:
+          path-to-signatures: "signatures/version1/cla.json"
+          path-to-document: "https://raw.githubusercontent.com/w-okada/voice-changer/master/LICENSE-CLA" # e.g. a CLA or a DCO document
+          # branch should not be protected
+          branch: "master"
+          #allowlist: user1,bot*
+
+          #below are the optional inputs - If the optional inputs are not given, then default values will be taken
+          #remote-organization-name: enter the remote organization name where the signatures should be stored (Default is storing the signatures in the same repository)
+          #remote-repository-name:  enter the  remote repository name where the signatures should be stored (Default is storing the signatures in the same repository)
+          #create-file-commit-message: 'For example: Creating file for storing CLA Signatures'
+          #signed-commit-message: 'For example: $contributorName has signed the CLA in #$pullRequestNo'
+          #custom-notsigned-prcomment: 'pull request comment with Introductory message to ask new contributors to sign'
+          #custom-pr-sign-comment: 'The signature to be committed in order to sign the CLA'
+          #custom-allsigned-prcomment: 'pull request comment when all contributors has signed, defaults to **CLA Assistant Lite bot** All Contributors have signed the CLA.'
+          #lock-pullrequest-aftermerge: false - if you don't want this bot to automatically lock the pull request after merging (default - true)
+          #use-dco-flag: true - If you are using DCO instead of CLA

LICENSE

@@ -20,7 +20,6 @@ LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 SOFTWARE.
 
-
 MIT License
 
 Copyright (c) 2022 Isle Tennos
@@ -63,4 +62,71 @@ FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
 AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
-SOFTWARE.
+SOFTWARE.
+
+MIT License
+
+Copyright (c) 2023 liujing04
+Copyright (c) 2023 源文雨
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
+
+MIT License
+
+Copyright (c) 2023 yxlllc
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
+
+MIT License
+
+Copyright (c) 2023 yxlllc
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

LICENSE-CLA

@@ -0,0 +1,27 @@
+Contributor License Agreement
+
+Copyright (c) 2022 Wataru Okada
+
+本契約は、当社とあなた（以下、"貢献者"とします）の間で締結され、貢献者が当社に対してソフトウェアプロジェクト（以下、"プロジェクト"とします）に対する貢献（以下、"貢献"とします）を提供する際の条件を定めます。
+
+1. 貢献者は、提供する貢献が、貢献者自身のオリジナルな作品であり、商標、著作権、特許、または他の知的財産権を侵害していないことを保証します。
+
+2. 貢献者は、貢献を当社に対して無償で提供し、当社はそれを無制限に使用、複製、修正、公開、配布、サブライセンスを付与し、またその販売する権利を得ることに同意します。
+
+3. 本契約が終了した場合でも、第 2 項で述べた権利は当社に留保されます。
+
+4. 当社は貢献者の貢献を受け入れる義務を負わず、また貢献者に一切の補償をする義務を負わないことに貢献者は同意します。
+
+5. 本契約は当社と貢献者双方の書面による合意により修正されることがあります。
+
+"This Agreement is made between our Company and you (hereinafter referred to as "Contributor") and outlines the terms under which you provide your Contributions (hereinafter referred to as "Contributions") to our software project (hereinafter referred to as "Project").
+
+1. You warrant that the Contributions you are providing are your original work and do not infringe any trademark, copyright, patent, or other intellectual property rights.
+
+2. You agree to provide your Contributions to the Company for free, and the Company has the unlimited right to use, copy, modify, publish, distribute, and sublicense, and also sell the Contributions.
+
+3. Even after the termination of this Agreement, the rights mentioned in the above clause will be retained by the Company.
+
+4. The Company is under no obligation to accept your Contributions or to compensate you in any way for them, and you agree to this.
+
+5. This Agreement may be modified by written agreement between the Company and the Contributor."

README.md

@@ -4,74 +4,19 @@
 
 ## What's New!
 
-- v.1.5.3.10b
+- v.1.5.3.12
-
-  - improve:
-    - logger
-  - bugfix:
-    - RMVPE:different device bug (not finding root caused yet)
-    - RVC: when loading sample model, useIndex issue
-
-- v.1.5.3.10a
-
-  - Improvement:
-    - launch sequence
-    - onnx export process
-    - error handling in client
-  - bugfix:
-    - RMVPE for mac
-
-- v.1.5.3.10
-
-  - New Feature
-    - Support Diffusion SVC(only combo model)
-    - System audio capture(only for win)
-    - Support RMVPE
-  - improvement
-    - directml: set device id
-  - some bugfixes:
-    - noise suppression2
-  - etc.
-
-- v.1.5.3.9a
-
-  - some improvements:
-    - keep f0 detector setting
-    - MMVC: max chunksize for onnx
-    - etc
-  - some bugfixs:
-    - RVC: crepe fail to estimate f0
-    - RVC: fallback from half-precision when half-precision failed.
-    - etc
-
-- v.1.5.3.9
-
-  - New feature:
-    - Add Crepe Full/Tiny (onnx)
-  - some improvements:
-    - server info includes python version
-    - contentvec onnx support
-    - etc
-  - some bugfixs:
-    - server device mode chuttering
-    - new model add sample rate
-    - etc
-
-- v.1.5.3.8a
-
-  - Bugfix(test): force client device samplerate
-  - Bugfix: server device filter
-
-- v.1.5.3.8
-
-  - RVC: performance improvement ([PR](https://github.com/w-okada/voice-changer/pull/371) from [nadare881](https://github.com/nadare881))
-
-- v.1.5.3.7
 
   - Feature:
-    - server device monitor
+    - Pass through mode
-  - Bugfix:
+  - bugfix:
-    - device output recorder button is showed in server device mode.
+    - Adapted the GUI to the number of slots.
+
+- v.1.5.3.11
+
+  - improve:
+    - increase slot size
+  - bugfix:
+    - m1 mac: eliminate torchaudio
 
 # VC Client とは
 
@@ -125,32 +70,14 @@
 
 - ダウンロードはこちらから。
 
-| Version     | OS  | フレームワーク                        | link                                                                                                                                                            | サポート VC                                                               | サイズ |
+| Version    | OS  | フレームワーク                        | link                                                                                                                                                            | サポート VC                                                               | サイズ |
-| ----------- | --- | ------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------- | ------ |
+| ---------- | --- | ------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------- | ------ |
-| v.1.5.3.10b | mac | ONNX(cpu), PyTorch(cpu,mps)           | [google](https://drive.google.com/uc?id=1akrb9RicU1-cldisToBedaM08y8pFQae&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, Diffusion-SVC           | 795MB  |
+| v.1.5.3.12 | mac | ONNX(cpu), PyTorch(cpu,mps)           | [google](https://drive.google.com/uc?id=1rC7IVpzfG68Ps6tBmdFIjSXvTNaUKBf6&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC                          | 797MB  |
-|             | win | ONNX(cpu,cuda), PyTorch(cpu,cuda)     | [google](https://drive.google.com/uc?id=1eZB0u2u0tEB1tR9mp06YiKx96x2oxgrN&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC, Diffusion-SVC | 3237MB |
+|            | win | ONNX(cpu,cuda), PyTorch(cpu,cuda)     | [google](https://drive.google.com/uc?id=1OqxS_jve4qvj71DdSGOrhI8DGaEVRzgs&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC, Diffusion-SVC | 3241MB |
-|             | win | ONNX(cpu,DirectML), PyTorch(cpu,cuda) | [google](https://drive.google.com/uc?id=1gzWuEN7oY_WdBwOEwtDdNaK2rT0nHvfN&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC, Diffusion-SVC | 3122MB |
+|            | win | ONNX(cpu,DirectML), PyTorch(cpu,cuda) | [google](https://drive.google.com/uc?id=1HhfmMovujzbOmvCi7WPuqQAuuo7jaM1o&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC, Diffusion-SVC | 3126MB |
-| v.1.5.3.10a | mac | ONNX(cpu), PyTorch(cpu,mps)           | [google](https://drive.google.com/uc?id=1_fLdFVswhOGwjRiQj4YWE-YZTO_GsnrA&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, Diffusion-SVC           | 795MB  |
+| v.1.5.3.11 | mac | ONNX(cpu), PyTorch(cpu,mps)           | [google](https://drive.google.com/uc?id=1cutPICJa-PI_ww0E3ae9FCuSjY_5PnWE&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC,                         | 795MB  |
-|             | win | ONNX(cpu,cuda), PyTorch(cpu,cuda)     | [google](https://drive.google.com/uc?id=1imaTBgWBb9ICkNy9pN6NxBISI6SzhEfL&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC, Diffusion-SVC | 3237MB |
+|            | win | ONNX(cpu,cuda), PyTorch(cpu,cuda)     | [google](https://drive.google.com/uc?id=1aOkc-QhtAj11gI8i335mHhNMUSESeJ5J&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC, Diffusion-SVC | 3237MB |
-|             | win | ONNX(cpu,DirectML), PyTorch(cpu,cuda) | [google](https://drive.google.com/uc?id=1GoijW29pjscdvxMhi8xgvPSvcHCGYwXO&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC, Diffusion-SVC | 3122MB |
+|            | win | ONNX(cpu,DirectML), PyTorch(cpu,cuda) | [google](https://drive.google.com/uc?id=16g33cZ925HNty_0Hly7Aw_nXlQlgqxDC&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC, Diffusion-SVC | 3122MB |
-| v.1.5.3.10  | mac | ONNX(cpu), PyTorch(cpu,mps)           | [google](https://drive.google.com/uc?id=1useZ4gcI0la5OhPuvt2j94CbAhWikpV4&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, Diffusion-SVC           | 795MB  |
-|             | win | ONNX(cpu,cuda), PyTorch(cpu,cuda)     | [google](https://drive.google.com/uc?id=13abR2xs4KmNIg9b5RJXFez9g6zwZqMj4&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC, Diffusion-SVC | 3237MB |
-|             | win | ONNX(cpu,DirectML), PyTorch(cpu,cuda) | [google](https://drive.google.com/uc?id=1ZxPp-HF7vSEJ8m00WnQaGbo4bTN4LqYD&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC, Diffusion-SVC | 3122MB |
-| v.1.5.3.9a  | mac | ONNX(cpu), PyTorch(cpu,mps)           | [google](https://drive.google.com/uc?id=1GsPTUTUbMvwNwAA8SGvSplwsf-yui0iw&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC                          | 794MB  |
-|             | win | ONNX(cpu,cuda), PyTorch(cpu,cuda)     | [google](https://drive.google.com/uc?id=1eKZCozh37QDfAr33ZG7lGFUOQv1tOooR&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC                | 3237MB |
-|             | win | ONNX(cpu,DirectML), PyTorch(cpu,cuda) | [google](https://drive.google.com/uc?id=1sxUNBPkeSPPNOE1ZknVF-0kx2jHP3kN6&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC                | 3122MB |
-| v.1.5.3.9   | mac | ONNX(cpu), PyTorch(cpu,mps)           | [google](https://drive.google.com/uc?id=1pTTcTseSdIfCyNUjB-K1mYPg9YocSYz6&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC                          | 795MB  |
-|             | win | ONNX(cpu,cuda), PyTorch(cpu,cuda)     | [google](https://drive.google.com/uc?id=1KWg-QoF6XmLbkUav-fmxc7bdAcD3844V&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC                | 3238MB |
-|             | win | ONNX(cpu,DirectML), PyTorch(cpu,cuda) | [google](https://drive.google.com/uc?id=1_TXUkDcofYz9mJd2L1ajAoyIBCQF29WL&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC                | 3123MB |
-| v.1.5.3.8a  | mac | ONNX(cpu), PyTorch(cpu,mps)           | [normal](https://drive.google.com/uc?id=1hg6lynE3wWJTNTParTa2qB2L06OL9KJ9&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC                          | 794MB  |
-|             | win | ONNX(cpu,DirectML), PyTorch(cpu,cuda) | [normal](https://drive.google.com/uc?id=1C9PCu8pdafO6jJ2yCaB7x54Ls7LcM0Xc&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC                | 3122MB |
-|             | win | ONNX(cpu,cuda), PyTorch(cpu,cuda)     | [normal](https://drive.google.com/uc?id=1bzrGhHPc9GdaRAMxkksTGtbuRLEeBx9i&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC                | 3237MB |
-| v.1.5.3.8   | mac | ONNX(cpu), PyTorch(cpu,mps)           | [normal](https://drive.google.com/uc?id=1ptmjFCRDW7M0l80072JVRII5tJpF13__&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC                          | 794MB  |
-|             | win | ONNX(cpu,DirectML), PyTorch(cpu,cuda) | [normal](https://drive.google.com/uc?id=19DfeACmpnzqCVH5bIoFunS2pGPABRuso&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC                | 3122MB |
-|             | win | ONNX(cpu,cuda), PyTorch(cpu,cuda)     | [normal](https://drive.google.com/uc?id=1AYP_hMdoeacX0KiF31Vd3oEjxwdreSbM&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC                | 3237MB |
-| v.1.5.3.7   | mac | ONNX(cpu), PyTorch(cpu,mps)           | [normal](https://drive.google.com/uc?id=1HdJwgo0__vR6pAkOkekejUZJ0lu2NfDs&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC                          | 794MB  |
-|             | win | ONNX(cpu,cuda), PyTorch(cpu,cuda)     | [normal](https://drive.google.com/uc?id=1JIF4PvKg-8HNUv_fMaXSM3AeYa-F_c4z&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC                | 3237MB |
-|             | win | ONNX(cpu,DirectML), PyTorch(cpu,cuda) | [normal](https://drive.google.com/uc?id=1cJzRHmD3vk6av0Dvwj3v9Ef5KUsQYhKv&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC                | 3122MB |
 
 (\*1) Google Drive からダウンロードできない方は[hugging_face](https://huggingface.co/wok000/vcclient000/tree/main)からダウンロードしてみてください
 (\*2) 開発者が AMD のグラフィックボードを持っていないので動作確認していません。onnxruntime-directml を同梱しただけのものです。
@@ -255,3 +182,7 @@ Github Pages 上で実行できるため、ブラウザのみあれば様々な
 |            | win | ONNX(cpu,cuda), PyTorch(cpu,cuda) | [normal](https://drive.google.com/uc?id=1tmTMJRRggS2Sb4goU-eHlRvUBR88RZDl&export=download) \*1 | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, so-vits-svc 4.0v2, RVC, DDSP-SVC | 2872MB |
 | v.1.5.3.1  | mac | ONNX(cpu), PyTorch(cpu,mps)       | [normal](https://drive.google.com/uc?id=1oswF72q_cQQeXhIn6W275qLnoBAmcrR_&export=download) \*1 | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC                              | 796MB  |
 |            | win | ONNX(cpu,cuda), PyTorch(cpu,cuda) | [normal](https://drive.google.com/uc?id=1AWjDhW4w2Uljp1-9P8YUJBZsIlnhkJX2&export=download) \*1 | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, so-vits-svc 4.0v2, RVC, DDSP-SVC | 2872MB |
+
+# For Contributor
+
+このリポジトリは[CLA](https://raw.githubusercontent.com/w-okada/voice-changer/master/LICENSE-CLA)を設定しています。

README_dev_en.md

@@ -52,6 +52,8 @@ $ python3 MMVCServerSIO.py -p 18888 --https true \
 
 ```
 
+Access with Browser (currently only chrome is supported), then you can see gui.
+
 2-1. Trouble shoot
 
 (1) OSError: PortAudio library not found

README_dev_ja.md

@@ -51,6 +51,8 @@ $ python3 MMVCServerSIO.py -p 18888 --https true \
     --samples samples.json
 ```
 
+ブラウザ(Chrome のみサポート)でアクセスすると画面が表示されます。
+
 2-1. トラブルシュート
 
 (1) OSError: PortAudio library not found

README_en.md

@@ -4,74 +4,19 @@
 
 ## What's New!
 
-- v.1.5.3.10b
+- v.1.5.3.12
-
-  - improve:
-    - logger
-  - bugfix:
-    - RMVPE:different device bug (not finding root caused yet)
-    - RVC: when loading sample model, useIndex issue
-
-- v.1.5.3.10a
-
-  - Improvement:
-    - launch sequence
-    - onnx export process
-    - error handling in client
-  - bugfix:
-    - RMVPE for mac
-
-- v.1.5.3.10
-
-  - New Feature
-    - Support Diffusion SVC(only combo model)
-    - System audio capture(only for win)
-    - Support RMVPE
-  - improvement
-    - directml: set device id
-  - some bugfixes:
-    - noise suppression2
-  - etc.
-
-- v.1.5.3.9a
-
-  - some improvements:
-    - keep f0 detector setting
-    - MMVC: max chunksize for onnx
-    - etc
-  - some bugfixs:
-    - RVC: crepe fail to estimate f0
-    - RVC: fallback from half-precision when half-precision failed.
-    - etc
-
-- v.1.5.3.9
-
-  - New feature:
-    - Add Crepe Full/Tiny (onnx)
-  - some improvements:
-    - server info includes python version
-    - contentvec onnx support
-    - etc
-  - some bugfixs:
-    - server device mode chuttering
-    - new model add sample rate
-    - etc
-
-- v.1.5.3.8a
-
-  - Bugfix(test): force client device samplerate
-  - Bugfix: server device filter
-
-- v.1.5.3.8
-
-  - RVC: performance improvement ([PR](https://github.com/w-okada/voice-changer/pull/371) from [nadare881](https://github.com/nadare881))
-
-- v.1.5.3.7
 
   - Feature:
-    - server device monitor
+    - Pass through mode
-  - Bugfix:
+  - bugfix:
-    - device output recorder button is showed in server device mode.
+    - Adapted the GUI to the number of slots.
+
+- v.1.5.3.11
+
+  - improve:
+    - increase slot size
+  - bugfix:
+    - m1 mac: eliminate torchaudio
 
 # What is VC Client
 
@@ -122,32 +67,14 @@ It can be used in two main ways, in order of difficulty:
 
 - Download (When you cannot download from google drive, try [hugging_face](https://huggingface.co/wok000/vcclient000/tree/main))
 
-| Version     | OS  | Framework                             | link                                                                                                                                                            | support VC                                                                | size   |
+| Version    | OS  | Framework                             | link                                                                                                                                                            | support VC                                                                | size   |
-| ----------- | --- | ------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------- | ------ |
+| ---------- | --- | ------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------- | ------ |
-| v.1.5.3.10b | mac | ONNX(cpu), PyTorch(cpu,mps)           | [google](https://drive.google.com/uc?id=1akrb9RicU1-cldisToBedaM08y8pFQae&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, Diffusion-SVC           | 795MB  |
+| v.1.5.3.12 | mac | ONNX(cpu), PyTorch(cpu,mps)           | [google](https://drive.google.com/uc?id=1rC7IVpzfG68Ps6tBmdFIjSXvTNaUKBf6&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC                          | 797MB  |
-|             | win | ONNX(cpu,cuda), PyTorch(cpu,cuda)     | [google](https://drive.google.com/uc?id=1eZB0u2u0tEB1tR9mp06YiKx96x2oxgrN&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC, Diffusion-SVC | 3237MB |
+|            | win | ONNX(cpu,cuda), PyTorch(cpu,cuda)     | [google](https://drive.google.com/uc?id=1OqxS_jve4qvj71DdSGOrhI8DGaEVRzgs&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC, Diffusion-SVC | 3241MB |
-|             | win | ONNX(cpu,DirectML), PyTorch(cpu,cuda) | [google](https://drive.google.com/uc?id=1gzWuEN7oY_WdBwOEwtDdNaK2rT0nHvfN&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC, Diffusion-SVC | 3122MB |
+|            | win | ONNX(cpu,DirectML), PyTorch(cpu,cuda) | [google](https://drive.google.com/uc?id=1HhfmMovujzbOmvCi7WPuqQAuuo7jaM1o&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC, Diffusion-SVC | 3126MB |
-| v.1.5.3.10a | mac | ONNX(cpu), PyTorch(cpu,mps)           | [google](https://drive.google.com/uc?id=1_fLdFVswhOGwjRiQj4YWE-YZTO_GsnrA&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, Diffusion-SVC           | 795MB  |
+| v.1.5.3.11 | mac | ONNX(cpu), PyTorch(cpu,mps)           | [google](https://drive.google.com/uc?id=1cutPICJa-PI_ww0E3ae9FCuSjY_5PnWE&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC,                         | 795MB  |
-|             | win | ONNX(cpu,cuda), PyTorch(cpu,cuda)     | [google](https://drive.google.com/uc?id=1imaTBgWBb9ICkNy9pN6NxBISI6SzhEfL&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC, Diffusion-SVC | 3237MB |
+|            | win | ONNX(cpu,cuda), PyTorch(cpu,cuda)     | [google](https://drive.google.com/uc?id=1aOkc-QhtAj11gI8i335mHhNMUSESeJ5J&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC, Diffusion-SVC | 3237MB |
-|             | win | ONNX(cpu,DirectML), PyTorch(cpu,cuda) | [google](https://drive.google.com/uc?id=1GoijW29pjscdvxMhi8xgvPSvcHCGYwXO&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC, Diffusion-SVC | 3122MB |
+|            | win | ONNX(cpu,DirectML), PyTorch(cpu,cuda) | [google](https://drive.google.com/uc?id=16g33cZ925HNty_0Hly7Aw_nXlQlgqxDC&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC, Diffusion-SVC | 3122MB |
-| v.1.5.3.10  | mac | ONNX(cpu), PyTorch(cpu,mps)           | [google](https://drive.google.com/uc?id=1useZ4gcI0la5OhPuvt2j94CbAhWikpV4&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, Diffusion-SVC           | 795MB  |
-|             | win | ONNX(cpu,cuda), PyTorch(cpu,cuda)     | [google](https://drive.google.com/uc?id=13abR2xs4KmNIg9b5RJXFez9g6zwZqMj4&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC, Diffusion-SVC | 3237MB |
-|             | win | ONNX(cpu,DirectML), PyTorch(cpu,cuda) | [google](https://drive.google.com/uc?id=1ZxPp-HF7vSEJ8m00WnQaGbo4bTN4LqYD&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC, Diffusion-SVC | 3122MB |
-| v.1.5.3.9a  | mac | ONNX(cpu), PyTorch(cpu,mps)           | [google](https://drive.google.com/uc?id=1GsPTUTUbMvwNwAA8SGvSplwsf-yui0iw&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC                          | 794MB  |
-|             | win | ONNX(cpu,cuda), PyTorch(cpu,cuda)     | [google](https://drive.google.com/uc?id=1eKZCozh37QDfAr33ZG7lGFUOQv1tOooR&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC                | 3237MB |
-|             | win | ONNX(cpu,DirectML), PyTorch(cpu,cuda) | [google](https://drive.google.com/uc?id=1sxUNBPkeSPPNOE1ZknVF-0kx2jHP3kN6&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC                | 3122MB |
-| v.1.5.3.9   | mac | ONNX(cpu), PyTorch(cpu,mps)           | [google](https://drive.google.com/uc?id=1pTTcTseSdIfCyNUjB-K1mYPg9YocSYz6&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC                          | 795MB  |
-|             | win | ONNX(cpu,cuda), PyTorch(cpu,cuda)     | [google](https://drive.google.com/uc?id=1KWg-QoF6XmLbkUav-fmxc7bdAcD3844V&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC                | 3238MB |
-|             | win | ONNX(cpu,DirectML), PyTorch(cpu,cuda) | [google](https://drive.google.com/uc?id=1_TXUkDcofYz9mJd2L1ajAoyIBCQF29WL&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC                | 3123MB |
-| v.1.5.3.8a  | mac | ONNX(cpu), PyTorch(cpu,mps)           | [normal](https://drive.google.com/uc?id=1hg6lynE3wWJTNTParTa2qB2L06OL9KJ9&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC                          | 794MB  |
-|             | win | ONNX(cpu,DirectML), PyTorch(cpu,cuda) | [normal](https://drive.google.com/uc?id=1C9PCu8pdafO6jJ2yCaB7x54Ls7LcM0Xc&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC                | 3122MB |
-|             | win | ONNX(cpu,cuda), PyTorch(cpu,cuda)     | [normal](https://drive.google.com/uc?id=1bzrGhHPc9GdaRAMxkksTGtbuRLEeBx9i&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC                | 3237MB |
-| v.1.5.3.8   | mac | ONNX(cpu), PyTorch(cpu,mps)           | [normal](https://drive.google.com/uc?id=1ptmjFCRDW7M0l80072JVRII5tJpF13__&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC                          | 794MB  |
-|             | win | ONNX(cpu,DirectML), PyTorch(cpu,cuda) | [normal](https://drive.google.com/uc?id=19DfeACmpnzqCVH5bIoFunS2pGPABRuso&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC                | 3122MB |
-|             | win | ONNX(cpu,cuda), PyTorch(cpu,cuda)     | [normal](https://drive.google.com/uc?id=1AYP_hMdoeacX0KiF31Vd3oEjxwdreSbM&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC                | 3237MB |
-| v.1.5.3.7   | mac | ONNX(cpu), PyTorch(cpu,mps)           | [normal](https://drive.google.com/uc?id=1HdJwgo0__vR6pAkOkekejUZJ0lu2NfDs&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC                          | 794MB  |
-|             | win | ONNX(cpu,cuda), PyTorch(cpu,cuda)     | [normal](https://drive.google.com/uc?id=1JIF4PvKg-8HNUv_fMaXSM3AeYa-F_c4z&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC                | 3237MB |
-|             | win | ONNX(cpu,DirectML), PyTorch(cpu,cuda) | [normal](https://drive.google.com/uc?id=1cJzRHmD3vk6av0Dvwj3v9Ef5KUsQYhKv&export=download), [hugging face](https://huggingface.co/wok000/vcclient000/tree/main) | MMVC v.1.5.x, MMVC v.1.3.x, so-vits-svc 4.0, RVC, DDSP-SVC                | 3122MB |
 
 (\*1) You can also download from [hugging_face](https://huggingface.co/wok000/vcclient000/tree/main)
 (\*2) The developer does not have an AMD graphics card, so it has not been tested. This package only includes onnxruntime-directml.

client/demo/dist/assets/gui_settings/edition_dml.txt

@@ -0,0 +1 @@
+onnxdirectML-cuda

client/demo/package.json

@@ -11,8 +11,8 @@
         "build:dev": "npm-run-all clean webpack:dev",
         "start": "webpack-dev-server  --config webpack.dev.js",
         "build:mod": "cd ../lib && npm run build:dev && cd - && cp -r ../lib/dist/* node_modules/@dannadori/voice-changer-client-js/dist/",
-        "build:mod_dos": "cd ../lib && npm run build:dev && cd ../demo && copy  ../lib/dist/index.js node_modules/@dannadori/voice-changer-client-js/dist/",
+        "build:mod_dos": "cd ../lib && npm run build:dev && cd ../demo && npm-run-all build:mod_copy",
-        "build:mod_dos2": "copy  ../lib/dist/index.js node_modules/@dannadori/voice-changer-client-js/dist/",
+        "build:mod_copy": "XCOPY ..\\lib\\dist\\* .\\node_modules\\@dannadori\\voice-changer-client-js\\dist\\* /s /e /h /y",
         "test": "echo \"Error: no test specified\" && exit 1"
     },
     "keywords": [
@@ -26,17 +26,17 @@
         "@babel/preset-env": "^7.22.9",
         "@babel/preset-react": "^7.22.5",
         "@babel/preset-typescript": "^7.22.5",
-        "@types/node": "^20.4.5",
+        "@types/node": "^20.4.6",
-        "@types/react": "^18.2.17",
+        "@types/react": "^18.2.18",
         "@types/react-dom": "^18.2.7",
         "autoprefixer": "^10.4.14",
         "babel-loader": "^9.1.3",
         "copy-webpack-plugin": "^11.0.0",
         "css-loader": "^6.8.1",
-        "eslint": "^8.45.0",
+        "eslint": "^8.46.0",
-        "eslint-config-prettier": "^8.8.0",
+        "eslint-config-prettier": "^8.9.0",
         "eslint-plugin-prettier": "^5.0.0",
-        "eslint-plugin-react": "^7.33.0",
+        "eslint-plugin-react": "^7.33.1",
         "eslint-webpack-plugin": "^4.0.1",
         "html-loader": "^4.2.0",
         "html-webpack-plugin": "^5.5.3",
@@ -54,11 +54,11 @@
         "webpack-dev-server": "^4.15.1"
     },
     "dependencies": {
-        "@dannadori/voice-changer-client-js": "^1.0.164",
+        "@dannadori/voice-changer-client-js": "^1.0.166",
-        "@fortawesome/fontawesome-svg-core": "^6.4.0",
+        "@fortawesome/fontawesome-svg-core": "^6.4.2",
-        "@fortawesome/free-brands-svg-icons": "^6.4.0",
+        "@fortawesome/free-brands-svg-icons": "^6.4.2",
-        "@fortawesome/free-regular-svg-icons": "^6.4.0",
+        "@fortawesome/free-regular-svg-icons": "^6.4.2",
-        "@fortawesome/free-solid-svg-icons": "^6.4.0",
+        "@fortawesome/free-solid-svg-icons": "^6.4.2",
         "@fortawesome/react-fontawesome": "^0.2.0",
         "protobufjs": "^7.2.4",
         "react": "^18.2.0",

client/demo/public/assets/gui_settings/edition_dml.txt

@@ -0,0 +1 @@
+onnxdirectML-cuda

client/demo/src/components/demo/001_GuiStateProvider.tsx

@@ -63,6 +63,7 @@ type GuiStateAndMethod = {
     outputAudioDeviceInfo: MediaDeviceInfo[];
     audioInputForGUI: string;
     audioOutputForGUI: string;
+    audioMonitorForGUI: string;
     fileInputEchoback: boolean | undefined;
     shareScreenEnabled: boolean;
     audioOutputForAnalyzer: string;
@@ -70,6 +71,7 @@ type GuiStateAndMethod = {
     setOutputAudioDeviceInfo: (val: MediaDeviceInfo[]) => void;
     setAudioInputForGUI: (val: string) => void;
     setAudioOutputForGUI: (val: string) => void;
+    setAudioMonitorForGUI: (val: string) => void;
     setFileInputEchoback: (val: boolean) => void;
     setShareScreenEnabled: (val: boolean) => void;
     setAudioOutputForAnalyzer: (val: string) => void;
@@ -106,6 +108,7 @@ export const GuiStateProvider = ({ children }: Props) => {
     const [outputAudioDeviceInfo, setOutputAudioDeviceInfo] = useState<MediaDeviceInfo[]>([]);
     const [audioInputForGUI, setAudioInputForGUI] = useState<string>("none");
     const [audioOutputForGUI, setAudioOutputForGUI] = useState<string>("none");
+    const [audioMonitorForGUI, setAudioMonitorForGUI] = useState<string>("none");
     const [fileInputEchoback, setFileInputEchoback] = useState<boolean>(false); //最初のmuteが有効になるように。undefined <-- ??? falseしておけばよさそう。undefinedだとwarningがでる。
     const [shareScreenEnabled, setShareScreenEnabled] = useState<boolean>(false);
     const [audioOutputForAnalyzer, setAudioOutputForAnalyzer] = useState<string>("default");
@@ -270,6 +273,7 @@ export const GuiStateProvider = ({ children }: Props) => {
         outputAudioDeviceInfo,
         audioInputForGUI,
         audioOutputForGUI,
+        audioMonitorForGUI,
         fileInputEchoback,
         shareScreenEnabled,
         audioOutputForAnalyzer,
@@ -277,6 +281,7 @@ export const GuiStateProvider = ({ children }: Props) => {
         setOutputAudioDeviceInfo,
         setAudioInputForGUI,
         setAudioOutputForGUI,
+        setAudioMonitorForGUI,
         setFileInputEchoback,
         setShareScreenEnabled,
         setAudioOutputForAnalyzer,

client/demo/src/components/demo/904-1_MainScreen.tsx

@@ -19,7 +19,7 @@ export const MainScreen = (props: MainScreenProps) => {
     const guiState = useGuiState();
     const messageBuilderState = useMessageBuilder();
     useMemo(() => {
-        messageBuilderState.setMessage(__filename, "change_icon", { ja: "アイコン変更", en: "chage icon" });
+        messageBuilderState.setMessage(__filename, "change_icon", { ja: "アイコン変更", en: "change icon" });
         messageBuilderState.setMessage(__filename, "rename", { ja: "リネーム", en: "rename" });
         messageBuilderState.setMessage(__filename, "download", { ja: "ダウンロード", en: "download" });
         messageBuilderState.setMessage(__filename, "terms_of_use", { ja: "利用規約", en: "terms of use" });
@@ -99,7 +99,7 @@ export const MainScreen = (props: MainScreenProps) => {
         const slotRow = serverSetting.serverSetting.modelSlots.map((x, index) => {
             // モデルのアイコン
             const generateIconArea = (slotIndex: number, iconUrl: string, tooltip: boolean) => {
-                const realIconUrl = iconUrl.length > 0 ? iconUrl : "/assets/icons/noimage.png";
+                const realIconUrl = iconUrl.length > 0 ? serverSetting.serverSetting.voiceChangerParams.model_dir + "/" + slotIndex + "/" + iconUrl.split(/[\/\\]/).pop() : "/assets/icons/noimage.png";
                 const iconDivClass = tooltip ? "tooltip" : "";
                 const iconClass = tooltip ? "model-slot-icon-pointable" : "model-slot-icon";
                 return (

client/demo/src/components/demo/905_MergeLabDialog.tsx

@@ -3,158 +3,182 @@ import { useGuiState } from "./001_GuiStateProvider";
 import { useAppState } from "../../001_provider/001_AppStateProvider";
 import { MergeElement, RVCModelSlot, RVCModelType, VoiceChangerType } from "@dannadori/voice-changer-client-js";
 
-
 export const MergeLabDialog = () => {
-    const guiState = useGuiState()
+    const guiState = useGuiState();
 
-    const { serverSetting } = useAppState()
+    const { serverSetting } = useAppState();
-    const [currentFilter, setCurrentFilter] = useState<string>("")
+    const [currentFilter, setCurrentFilter] = useState<string>("");
-    const [mergeElements, setMergeElements] = useState<MergeElement[]>([])
+    const [mergeElements, setMergeElements] = useState<MergeElement[]>([]);
 
     // スロットが変更されたときの初期化処理
     const newSlotChangeKey = useMemo(() => {
         if (!serverSetting.serverSetting.modelSlots) {
-            return ""
+            return "";
         }
         return serverSetting.serverSetting.modelSlots.reduce((prev, cur) => {
-            return prev + "_" + cur.modelFile
+            return prev + "_" + cur.modelFile;
-        }, "")
+        }, "");
-    }, [serverSetting.serverSetting.modelSlots])
+    }, [serverSetting.serverSetting.modelSlots]);
 
     const filterItems = useMemo(() => {
-        return serverSetting.serverSetting.modelSlots.reduce((prev, cur) => {
+        return serverSetting.serverSetting.modelSlots.reduce(
-            if (cur.voiceChangerType != "RVC") {
+            (prev, cur) => {
-                return prev
+                if (cur.voiceChangerType != "RVC") {
-            }
+                    return prev;
-            const curRVC = cur as RVCModelSlot
+                }
-            const key = `${curRVC.modelType},${cur.samplingRate},${curRVC.embChannels}`
+                const curRVC = cur as RVCModelSlot;
-            const val = { type: curRVC.modelType, samplingRate: cur.samplingRate, embChannels: curRVC.embChannels }
+                const key = `${curRVC.modelType},${cur.samplingRate},${curRVC.embChannels}`;
-            const existKeys = Object.keys(prev)
+                const val = { type: curRVC.modelType, samplingRate: cur.samplingRate, embChannels: curRVC.embChannels };
-            if (!cur.modelFile || cur.modelFile.length == 0) {
+                const existKeys = Object.keys(prev);
-                return prev
+                if (!cur.modelFile || cur.modelFile.length == 0) {
-            }
+                    return prev;
-            if (curRVC.modelType == "onnxRVC" || curRVC.modelType == "onnxRVCNono") {
+                }
-                return prev
+                if (curRVC.modelType == "onnxRVC" || curRVC.modelType == "onnxRVCNono") {
-            }
+                    return prev;
-            if (!existKeys.includes(key)) {
+                }
-                prev[key] = val
+                if (!existKeys.includes(key)) {
-            }
+                    prev[key] = val;
-            return prev
+                }
-        }, {} as { [key: string]: { type: RVCModelType, samplingRate: number, embChannels: number } })
+                return prev;
-
+            },
-    }, [newSlotChangeKey])
+            {} as { [key: string]: { type: RVCModelType; samplingRate: number; embChannels: number } },
+        );
+    }, [newSlotChangeKey]);
 
     const models = useMemo(() => {
-        return serverSetting.serverSetting.modelSlots.filter(x => {
+        return serverSetting.serverSetting.modelSlots.filter((x) => {
             if (x.voiceChangerType != "RVC") {
-                return
+                return;
             }
-            const xRVC = x as RVCModelSlot
+            const xRVC = x as RVCModelSlot;
-            const filterVals = filterItems[currentFilter]
+            const filterVals = filterItems[currentFilter];
             if (!filterVals) {
-                return false
+                return false;
             }
             if (xRVC.modelType == filterVals.type && xRVC.samplingRate == filterVals.samplingRate && xRVC.embChannels == filterVals.embChannels) {
-                return true
+                return true;
             } else {
-                return false
+                return false;
             }
-        })
+        });
-    }, [filterItems, currentFilter])
+    }, [filterItems, currentFilter]);
 
     useEffect(() => {
         if (Object.keys(filterItems).length > 0) {
-            setCurrentFilter(Object.keys(filterItems)[0])
+            setCurrentFilter(Object.keys(filterItems)[0]);
         }
-    }, [filterItems])
+    }, [filterItems]);
     useEffect(() => {
+        // models はフィルタ後の配列
         const newMergeElements = models.map((x) => {
-            return { filename: x.modelFile, strength: 0 }
+            return { slotIndex: x.slotIndex, filename: x.modelFile, strength: 0 };
-        })
+        });
-        setMergeElements(newMergeElements)
+        setMergeElements(newMergeElements);
-    }, [models])
+    }, [models]);
 
     const dialog = useMemo(() => {
         const closeButtonRow = (
             <div className="body-row split-3-4-3 left-padding-1">
-                <div className="body-item-text">
+                <div className="body-item-text"></div>
-                </div>
                 <div className="body-button-container body-button-container-space-around">
-                    <div className="body-button" onClick={() => { guiState.stateControls.showMergeLabCheckbox.updateState(false) }} >close</div>
+                    <div
+                        className="body-button"
+                        onClick={() => {
+                            guiState.stateControls.showMergeLabCheckbox.updateState(false);
+                        }}
+                    >
+                        close
+                    </div>
                 </div>
                 <div className="body-item-text"></div>
             </div>
-        )
+        );
 
-
+        const filterOptions = Object.keys(filterItems)
-        const filterOptions = Object.keys(filterItems).map(x => {
+            .map((x) => {
-            return <option key={x} value={x}>{x}</option>
+                return (
-        }).filter(x => x != null)
+                    <option key={x} value={x}>
-
+                        {x}
-        const onMergeElementsChanged = (filename: string, strength: number) => {
+                    </option>
-            const newMergeElements = mergeElements.map((x) => {
+                );
-                if (x.filename == filename) {
-                    return { filename: x.filename, strength: strength }
-                } else {
-                    return x
-                }
             })
-            setMergeElements(newMergeElements)
+            .filter((x) => x != null);
-        }
+
+        const onMergeElementsChanged = (slotIndex: number, strength: number) => {
+            const newMergeElements = mergeElements.map((x) => {
+                if (x.slotIndex == slotIndex) {
+                    return { slotIndex: x.slotIndex, strength: strength };
+                } else {
+                    return x;
+                }
+            });
+            setMergeElements(newMergeElements);
+        };
 
         const onMergeClicked = () => {
+            const validMergeElements = mergeElements.filter((x) => {
+                return x.strength > 0;
+            });
             serverSetting.mergeModel({
                 voiceChangerType: VoiceChangerType.RVC,
                 command: "mix",
-                files: mergeElements
+                files: validMergeElements,
-            })
+            });
-        }
+        };
 
         const modelList = mergeElements.map((x, index) => {
-            const name = models.find(model => { return model.modelFile == x.filename })?.name || ""
+            const name =
+                models.find((model) => {
+                    return model.slotIndex == x.slotIndex;
+                })?.name || "";
+
             return (
                 <div key={index} className="merge-lab-model-item">
+                    <div>{name}</div>
                     <div>
-                        {name}
+                        <input
-                    </div>
+                            type="range"
-                    <div>
+                            className="body-item-input-slider"
-                        <input type="range" className="body-item-input-slider" min="0" max="100" step="1" value={x.strength} onChange={(e) => {
+                            min="0"
-                            onMergeElementsChanged(x.filename, Number(e.target.value))
+                            max="100"
-                        }}></input>
+                            step="1"
+                            value={x.strength}
+                            onChange={(e) => {
+                                onMergeElementsChanged(x.slotIndex, Number(e.target.value));
+                            }}
+                        ></input>
                         <span className="body-item-input-slider-val">{x.strength}</span>
                     </div>
                 </div>
-            )
+            );
-        })
+        });
-
 
         const content = (
             <div className="merge-lab-container">
                 <div className="merge-lab-type-filter">
+                    <div>Type:</div>
                     <div>
-                        Type:
+                        <select
-                    </div>
+                            value={currentFilter}
-                    <div>
+                            onChange={(e) => {
-                        <select value={currentFilter} onChange={(e) => { setCurrentFilter(e.target.value) }}>
+                                setCurrentFilter(e.target.value);
+                            }}
+                        >
                             {filterOptions}
                         </select>
                     </div>
                 </div>
                 <div className="merge-lab-manipulator">
-                    <div className="merge-lab-model-list">
+                    <div className="merge-lab-model-list">{modelList}</div>
-                        {modelList}
-                    </div>
                     <div className="merge-lab-merge-buttons">
-                        <div className="merge-lab-merge-buttons-notice">
+                        <div className="merge-lab-merge-buttons-notice">The merged model is stored in the final slot. If you assign this slot, it will be overwritten.</div>
-                            The merged model is stored in the final slot. If you assign this slot, it will be overwritten.
-                        </div>
                         <div className="merge-lab-merge-button" onClick={onMergeClicked}>
                             merge
                         </div>
                     </div>
                 </div>
             </div>
-        )
+        );
         return (
             <div className="dialog-frame">
                 <div className="dialog-title">MergeLab</div>
@@ -166,5 +190,4 @@ export const MergeLabDialog = () => {
         );
     }, [newSlotChangeKey, currentFilter, mergeElements, models]);
     return dialog;
-
 };

client/demo/src/components/demo/components2/100_ModelSlotArea.tsx

@@ -1,83 +1,115 @@
-import React, { useMemo } from "react"
+import React, { useMemo, useState } from "react";
-import { useAppState } from "../../../001_provider/001_AppStateProvider"
+import { useAppState } from "../../../001_provider/001_AppStateProvider";
-import { useGuiState } from "../001_GuiStateProvider"
+import { useGuiState } from "../001_GuiStateProvider";
-import { useMessageBuilder } from "../../../hooks/useMessageBuilder"
+import { useMessageBuilder } from "../../../hooks/useMessageBuilder";
+import { FontAwesomeIcon } from "@fortawesome/react-fontawesome";
 
-export type ModelSlotAreaProps = {
+export type ModelSlotAreaProps = {};
-}
 
+const SortTypes = {
+    slot: "slot",
+    name: "name",
+} as const;
+export type SortTypes = (typeof SortTypes)[keyof typeof SortTypes];
 
 export const ModelSlotArea = (_props: ModelSlotAreaProps) => {
-    const { serverSetting, getInfo } = useAppState()
+    const { serverSetting, getInfo } = useAppState();
-    const guiState = useGuiState()
+    const guiState = useGuiState();
-    const messageBuilderState = useMessageBuilder()
+    const messageBuilderState = useMessageBuilder();
+    const [sortType, setSortType] = useState<SortTypes>("slot");
 
     useMemo(() => {
-        messageBuilderState.setMessage(__filename, "edit", { "ja": "編集", "en": "edit" })
+        messageBuilderState.setMessage(__filename, "edit", { ja: "編集", en: "edit" });
-    }, [])
+    }, []);
-
 
     const modelTiles = useMemo(() => {
         if (!serverSetting.serverSetting.modelSlots) {
-            return []
+            return [];
         }
-        return serverSetting.serverSetting.modelSlots.map((x, index) => {
+        const modelSlots =
-            if (!x.modelFile || x.modelFile.length == 0) {
+            sortType == "slot"
-                return null
+                ? serverSetting.serverSetting.modelSlots
-            }
+                : serverSetting.serverSetting.modelSlots.slice().sort((a, b) => {
-            const tileContainerClass = index == serverSetting.serverSetting.modelSlotIndex ? "model-slot-tile-container-selected" : "model-slot-tile-container"
+                      return a.name.localeCompare(b.name);
-            const name = x.name.length > 8 ? x.name.substring(0, 7) + "..." : x.name
+                  });
-            const iconElem = x.iconFile.length > 0 ?
-                <>
-                    <img className="model-slot-tile-icon" src={x.iconFile} alt={x.name} />
-                    <div className="model-slot-tile-vctype">{x.voiceChangerType}</div>
-                </>
-                :
-                <>
-                    <div className="model-slot-tile-icon-no-entry">no image</div>
-                    <div className="model-slot-tile-vctype">{x.voiceChangerType}</div>
-                </>
 
-            const clickAction = async () => {
+        return modelSlots
-                const dummyModelSlotIndex = (Math.floor(Date.now() / 1000)) * 1000 + index
+            .map((x, index) => {
-                await serverSetting.updateServerSettings({ ...serverSetting.serverSetting, modelSlotIndex: dummyModelSlotIndex })
+                if (!x.modelFile || x.modelFile.length == 0) {
-                setTimeout(() => { // quick hack
+                    return null;
-                    getInfo()
+                }
-                }, 1000 * 2)
+                const tileContainerClass = x.slotIndex == serverSetting.serverSetting.modelSlotIndex ? "model-slot-tile-container-selected" : "model-slot-tile-container";
-            }
+                const name = x.name.length > 8 ? x.name.substring(0, 7) + "..." : x.name;
 
-            return (
+                const iconElem =
-                <div key={index} className={tileContainerClass} onClick={clickAction}>
+                    x.iconFile.length > 0 ? (
-                    <div className="model-slot-tile-icon-div">
+                        <>
-                        {iconElem}
+                            <img className="model-slot-tile-icon" src={serverSetting.serverSetting.voiceChangerParams.model_dir + "/" + x.slotIndex + "/" + x.iconFile.split(/[\/\\]/).pop()} alt={x.name} />
+                            <div className="model-slot-tile-vctype">{x.voiceChangerType}</div>
+                        </>
+                    ) : (
+                        <>
+                            <div className="model-slot-tile-icon-no-entry">no image</div>
+                            <div className="model-slot-tile-vctype">{x.voiceChangerType}</div>
+                        </>
+                    );
+
+                const clickAction = async () => {
+                    const dummyModelSlotIndex = Math.floor(Date.now() / 1000) * 1000 + x.slotIndex;
+                    await serverSetting.updateServerSettings({ ...serverSetting.serverSetting, modelSlotIndex: dummyModelSlotIndex });
+                    setTimeout(() => {
+                        // quick hack
+                        getInfo();
+                    }, 1000 * 2);
+                };
+
+                return (
+                    <div key={index} className={tileContainerClass} onClick={clickAction}>
+                        <div className="model-slot-tile-icon-div">{iconElem}</div>
+                        <div className="model-slot-tile-dscription">{name}</div>
                     </div>
-                    <div className="model-slot-tile-dscription">
+                );
-                        {name}
+            })
-                    </div>
+            .filter((x) => x != null);
-                </div >
+    }, [serverSetting.serverSetting.modelSlots, serverSetting.serverSetting.modelSlotIndex, sortType]);
-            )
-        }).filter(x => x != null)
-    }, [serverSetting.serverSetting.modelSlots, serverSetting.serverSetting.modelSlotIndex])
-
 
     const modelSlotArea = useMemo(() => {
         const onModelSlotEditClicked = () => {
-            guiState.stateControls.showModelSlotManagerCheckbox.updateState(true)
+            guiState.stateControls.showModelSlotManagerCheckbox.updateState(true);
-        }
+        };
+        const sortSlotByIdClass = sortType == "slot" ? "model-slot-sort-button-active" : "model-slot-sort-button";
+        const sortSlotByNameClass = sortType == "name" ? "model-slot-sort-button-active" : "model-slot-sort-button";
         return (
             <div className="model-slot-area">
                 <div className="model-slot-panel">
                     <div className="model-slot-tiles-container">{modelTiles}</div>
                     <div className="model-slot-buttons">
+                        <div className="model-slot-sort-buttons">
+                            <div
+                                className={sortSlotByIdClass}
+                                onClick={() => {
+                                    setSortType("slot");
+                                }}
+                            >
+                                <FontAwesomeIcon icon={["fas", "arrow-down-1-9"]} style={{ fontSize: "1rem" }} />
+                            </div>
+                            <div
+                                className={sortSlotByNameClass}
+                                onClick={() => {
+                                    setSortType("name");
+                                }}
+                            >
+                                <FontAwesomeIcon icon={["fas", "arrow-down-a-z"]} style={{ fontSize: "1rem" }} />
+                            </div>
+                        </div>
                         <div className="model-slot-button" onClick={onModelSlotEditClicked}>
                             {messageBuilderState.getMessage(__filename, "edit")}
                         </div>
                     </div>
-
                 </div>
             </div>
-        )
+        );
-    }, [modelTiles])
+    }, [modelTiles, sortType]);
 
-    return modelSlotArea
+    return modelSlotArea;
-}
+};

client/demo/src/components/demo/components2/101-7_diffusion-svcSettingArea.tsx

@@ -23,6 +23,26 @@ export const DiffusionSVCSettingArea = (_props: DiffusionSVCSettingAreaProps) =>
             return <></>;
         }
 
+        const skipDiffusionClass = serverSetting.serverSetting.skipDiffusion == 0 ? "character-area-toggle-button" : "character-area-toggle-button-active";
+
+        const skipDiffRow = (
+            <div className="character-area-control">
+                <div className="character-area-control-title">Boost</div>
+                <div className="character-area-control-field">
+                    <div className="character-area-buttons">
+                        <div
+                            className={skipDiffusionClass}
+                            onClick={() => {
+                                serverSetting.updateServerSettings({ ...serverSetting.serverSetting, skipDiffusion: serverSetting.serverSetting.skipDiffusion == 0 ? 1 : 0 });
+                            }}
+                        >
+                            skip diff
+                        </div>
+                    </div>
+                </div>
+            </div>
+        );
+
         const skipValues = getDivisors(serverSetting.serverSetting.kStep);
         skipValues.pop();
 
@@ -82,6 +102,7 @@ export const DiffusionSVCSettingArea = (_props: DiffusionSVCSettingAreaProps) =>
         );
         return (
             <>
+                {skipDiffRow}
                 {kStepRow}
                 {speedUpRow}
             </>

client/demo/src/components/demo/components2/101_CharacterArea.tsx

@@ -49,7 +49,7 @@ export const CharacterArea = (_props: CharacterAreaProps) => {
             return <></>;
         }
 
-        const icon = selected.iconFile.length > 0 ? selected.iconFile : "./assets/icons/human.png";
+        const icon = selected.iconFile.length > 0 ? serverSetting.serverSetting.voiceChangerParams.model_dir + "/" + selected.slotIndex + "/" + selected.iconFile.split(/[\/\\]/).pop() : "./assets/icons/human.png";
         const selectedTermOfUseUrlLink = selected.termsOfUseUrl ? (
             <a href={selected.termsOfUseUrl} target="_blank" rel="noopener noreferrer" className="portrait-area-terms-of-use-link">
                 [{messageBuilderState.getMessage(__filename, "terms_of_use")}]
@@ -122,9 +122,13 @@ export const CharacterArea = (_props: CharacterAreaProps) => {
                 serverSetting.updateServerSettings({ ...serverSetting.serverSetting, serverAudioStated: 0 });
             }
         };
+        const onPassThroughClicked = async () => {
+            serverSetting.updateServerSettings({ ...serverSetting.serverSetting, passThrough: !serverSetting.serverSetting.passThrough });
+        };
         const startClassName = guiState.isConverting ? "character-area-control-button-active" : "character-area-control-button-stanby";
         const stopClassName = guiState.isConverting ? "character-area-control-button-stanby" : "character-area-control-button-active";
-
+        const passThruClassName = serverSetting.serverSetting.passThrough == false ? "character-area-control-passthru-button-stanby" : "character-area-control-passthru-button-active blinking";
+        console.log("serverSetting.serverSetting.passThrough", passThruClassName, serverSetting.serverSetting.passThrough);
         return (
             <div className="character-area-control">
                 <div className="character-area-control-buttons">
@@ -134,6 +138,9 @@ export const CharacterArea = (_props: CharacterAreaProps) => {
                     <div onClick={onStopClicked} className={stopClassName}>
                         stop
                     </div>
+                    <div onClick={onPassThroughClicked} className={passThruClassName}>
+                        passthru
+                    </div>
                 </div>
             </div>
         );

client/demo/src/components/demo/components2/102-2_ConvertArea.tsx

@@ -41,68 +41,75 @@ export const ConvertArea = (props: ConvertProps) => {
 
         const gpuSelect =
             edition.indexOf("onnxdirectML-cuda") >= 0 ? (
-                <div className="config-sub-area-control">
+                <>
-                    <div className="config-sub-area-control-title">GPU(dml):</div>
+                    <div className="config-sub-area-control">
-                    <div className="config-sub-area-control-field">
+                        <div className="config-sub-area-control-title">GPU(dml):</div>
-                        <div className="config-sub-area-buttons">
+                        <div className="config-sub-area-control-field">
-                            <div
+                            <div className="config-sub-area-buttons">
-                                onClick={async () => {
+                                <div
-                                    await serverSetting.updateServerSettings({
+                                    onClick={async () => {
-                                        ...serverSetting.serverSetting,
+                                        await serverSetting.updateServerSettings({
-                                        gpu: -1,
+                                            ...serverSetting.serverSetting,
-                                    });
+                                            gpu: -1,
-                                }}
+                                        });
-                                className={cpuClassName}
+                                    }}
-                            >
+                                    className={cpuClassName}
-                                cpu
+                                >
-                            </div>
+                                    <span className="config-sub-area-button-text-small">cpu</span>
-                            <div
+                                </div>
-                                onClick={async () => {
+                                <div
-                                    await serverSetting.updateServerSettings({
+                                    onClick={async () => {
-                                        ...serverSetting.serverSetting,
+                                        await serverSetting.updateServerSettings({
-                                        gpu: 0,
+                                            ...serverSetting.serverSetting,
-                                    });
+                                            gpu: 0,
-                                }}
+                                        });
-                                className={gpu0ClassName}
+                                    }}
-                            >
+                                    className={gpu0ClassName}
-                                0
+                                >
-                            </div>
+                                    <span className="config-sub-area-button-text-small">gpu0</span>
-                            <div
+                                </div>
-                                onClick={async () => {
+                                <div
-                                    await serverSetting.updateServerSettings({
+                                    onClick={async () => {
-                                        ...serverSetting.serverSetting,
+                                        await serverSetting.updateServerSettings({
-                                        gpu: 1,
+                                            ...serverSetting.serverSetting,
-                                    });
+                                            gpu: 1,
-                                }}
+                                        });
-                                className={gpu1ClassName}
+                                    }}
-                            >
+                                    className={gpu1ClassName}
-                                1
+                                >
-                            </div>
+                                    <span className="config-sub-area-button-text-small">gpu1</span>
-                            <div
+                                </div>
-                                onClick={async () => {
+                                <div
-                                    await serverSetting.updateServerSettings({
+                                    onClick={async () => {
-                                        ...serverSetting.serverSetting,
+                                        await serverSetting.updateServerSettings({
-                                        gpu: 2,
+                                            ...serverSetting.serverSetting,
-                                    });
+                                            gpu: 2,
-                                }}
+                                        });
-                                className={gpu2ClassName}
+                                    }}
-                            >
+                                    className={gpu2ClassName}
-                                2
+                                >
-                            </div>
+                                    <span className="config-sub-area-button-text-small">gpu2</span>
-                            <div
+                                </div>
-                                onClick={async () => {
+                                <div
-                                    await serverSetting.updateServerSettings({
+                                    onClick={async () => {
-                                        ...serverSetting.serverSetting,
+                                        await serverSetting.updateServerSettings({
-                                        gpu: 3,
+                                            ...serverSetting.serverSetting,
-                                    });
+                                            gpu: 3,
-                                }}
+                                        });
-                                className={gpu3ClassName}
+                                    }}
-                            >
+                                    className={gpu3ClassName}
-                                3
+                                >
+                                    <span className="config-sub-area-button-text-small">gpu3</span>
+                                </div>
+                                <div className="config-sub-area-control">
+                                    <span className="config-sub-area-button-text-small">
+                                        <a href="https://github.com/w-okada/voice-changer/issues/410">more info</a>
+                                    </span>
+                                </div>
                             </div>
                         </div>
                     </div>
-                </div>
+                </>
             ) : (
                 <div className="config-sub-area-control">
                     <div className="config-sub-area-control-title">GPU:</div>

client/demo/src/components/demo/components2/102-3_DeviceArea.tsx

@@ -2,14 +2,14 @@ import React, { useEffect, useMemo, useRef, useState } from "react";
 import { useAppState } from "../../../001_provider/001_AppStateProvider";
 import { fileSelectorAsDataURL, useIndexedDB } from "@dannadori/voice-changer-client-js";
 import { useGuiState } from "../001_GuiStateProvider";
-import { AUDIO_ELEMENT_FOR_PLAY_RESULT, AUDIO_ELEMENT_FOR_TEST_CONVERTED, AUDIO_ELEMENT_FOR_TEST_CONVERTED_ECHOBACK, AUDIO_ELEMENT_FOR_TEST_ORIGINAL, INDEXEDDB_KEY_AUDIO_OUTPUT } from "../../../const";
+import { AUDIO_ELEMENT_FOR_PLAY_MONITOR, AUDIO_ELEMENT_FOR_PLAY_RESULT, AUDIO_ELEMENT_FOR_TEST_CONVERTED, AUDIO_ELEMENT_FOR_TEST_CONVERTED_ECHOBACK, AUDIO_ELEMENT_FOR_TEST_ORIGINAL, INDEXEDDB_KEY_AUDIO_MONITR, INDEXEDDB_KEY_AUDIO_OUTPUT } from "../../../const";
 import { isDesktopApp } from "../../../const";
 
 export type DeviceAreaProps = {};
 
 export const DeviceArea = (_props: DeviceAreaProps) => {
-    const { setting, serverSetting, audioContext, setAudioOutputElementId, initializedRef, setVoiceChangerClientSetting, startOutputRecording, stopOutputRecording } = useAppState();
+    const { setting, serverSetting, audioContext, setAudioOutputElementId, setAudioMonitorElementId, initializedRef, setVoiceChangerClientSetting, startOutputRecording, stopOutputRecording } = useAppState();
-    const { isConverting, audioInputForGUI, inputAudioDeviceInfo, setAudioInputForGUI, fileInputEchoback, setFileInputEchoback, setAudioOutputForGUI, audioOutputForGUI, outputAudioDeviceInfo, shareScreenEnabled, setShareScreenEnabled } = useGuiState();
+    const { isConverting, audioInputForGUI, inputAudioDeviceInfo, setAudioInputForGUI, fileInputEchoback, setFileInputEchoback, setAudioOutputForGUI, setAudioMonitorForGUI, audioOutputForGUI, audioMonitorForGUI, outputAudioDeviceInfo, shareScreenEnabled, setShareScreenEnabled } = useGuiState();
     const [inputHostApi, setInputHostApi] = useState<string>("ALL");
     const [outputHostApi, setOutputHostApi] = useState<string>("ALL");
     const [monitorHostApi, setMonitorHostApi] = useState<string>("ALL");
@@ -244,10 +244,10 @@ export const DeviceArea = (_props: DeviceAreaProps) => {
             audio_echo.volume = 0;
             setFileInputEchoback(false);
 
-            // original stream to play.
+            // // original stream to play.
-            const audio_org = document.getElementById(AUDIO_ELEMENT_FOR_TEST_ORIGINAL) as HTMLAudioElement;
+            // const audio_org = document.getElementById(AUDIO_ELEMENT_FOR_TEST_ORIGINAL) as HTMLAudioElement;
-            audio_org.src = url;
+            // audio_org.src = url;
-            audio_org.pause();
+            // audio_org.pause();
         };
 
         const echobackClass = fileInputEchoback ? "config-sub-area-control-field-wav-file-echoback-button-active" : "config-sub-area-control-field-wav-file-echoback-button";
@@ -256,7 +256,7 @@ export const DeviceArea = (_props: DeviceAreaProps) => {
                 <div className="config-sub-area-control-field">
                     <div className="config-sub-area-control-field-wav-file left-padding-1">
                         <div className="config-sub-area-control-field-wav-file-audio-container">
-                            <audio id={AUDIO_ELEMENT_FOR_TEST_ORIGINAL} controls hidden></audio>
+                            {/* <audio id={AUDIO_ELEMENT_FOR_TEST_ORIGINAL} controls hidden></audio> */}
                             <audio className="config-sub-area-control-field-wav-file-audio" id={AUDIO_ELEMENT_FOR_TEST_CONVERTED} controls controlsList="nodownload noplaybackrate"></audio>
                             <audio id={AUDIO_ELEMENT_FOR_TEST_CONVERTED_ECHOBACK} controls hidden></audio>
                         </div>
@@ -381,7 +381,8 @@ export const DeviceArea = (_props: DeviceAreaProps) => {
         const setAudioOutput = async () => {
             const mediaDeviceInfos = await navigator.mediaDevices.enumerateDevices();
 
-            [AUDIO_ELEMENT_FOR_PLAY_RESULT, AUDIO_ELEMENT_FOR_TEST_ORIGINAL, AUDIO_ELEMENT_FOR_TEST_CONVERTED_ECHOBACK].forEach((x) => {
+            // [AUDIO_ELEMENT_FOR_PLAY_RESULT, AUDIO_ELEMENT_FOR_TEST_ORIGINAL, AUDIO_ELEMENT_FOR_TEST_CONVERTED_ECHOBACK].forEach((x) => {
+            [AUDIO_ELEMENT_FOR_PLAY_RESULT, AUDIO_ELEMENT_FOR_TEST_CONVERTED_ECHOBACK].forEach((x) => {
                 const audio = document.getElementById(x) as HTMLAudioElement;
                 if (audio) {
                     if (serverSetting.serverSetting.enableServerAudio == 1) {
@@ -598,7 +599,88 @@ export const DeviceArea = (_props: DeviceAreaProps) => {
         );
     }, [serverSetting.serverSetting, serverSetting.updateServerSettings, serverSetting.serverSetting.enableServerAudio]);
 
-    // (6) Monitor
+    // (6) モニター
+    useEffect(() => {
+        const loadCache = async () => {
+            const key = await getItem(INDEXEDDB_KEY_AUDIO_MONITR);
+            if (key) {
+                setAudioMonitorForGUI(key as string);
+            }
+        };
+        loadCache();
+    }, []);
+    useEffect(() => {
+        const setAudioMonitor = async () => {
+            const mediaDeviceInfos = await navigator.mediaDevices.enumerateDevices();
+
+            [AUDIO_ELEMENT_FOR_PLAY_MONITOR].forEach((x) => {
+                const audio = document.getElementById(x) as HTMLAudioElement;
+                if (audio) {
+                    if (serverSetting.serverSetting.enableServerAudio == 1) {
+                        // Server Audio を使う場合はElementから音は出さない。
+                        audio.volume = 0;
+                    } else if (audioMonitorForGUI == "none") {
+                        // @ts-ignore
+                        audio.setSinkId("");
+                        audio.volume = 0;
+                    } else {
+                        const audioOutputs = mediaDeviceInfos.filter((x) => {
+                            return x.kind == "audiooutput";
+                        });
+                        const found = audioOutputs.some((x) => {
+                            return x.deviceId == audioMonitorForGUI;
+                        });
+                        if (found) {
+                            // @ts-ignore // 例外キャッチできないので事前にIDチェックが必要らしい。！？
+                            audio.setSinkId(audioMonitorForGUI);
+                            audio.volume = 1;
+                        } else {
+                            console.warn("No audio output device. use default");
+                        }
+                    }
+                }
+            });
+        };
+        setAudioMonitor();
+    }, [audioMonitorForGUI, serverSetting.serverSetting.enableServerAudio]);
+
+    // (6-1) クライアント
+    const clientMonitorRow = useMemo(() => {
+        if (serverSetting.serverSetting.enableServerAudio == 1) {
+            return <></>;
+        }
+
+        return (
+            <div className="config-sub-area-control">
+                <div className="config-sub-area-control-title left-padding-1">monitor</div>
+                <div className="config-sub-area-control-field">
+                    <select
+                        className="body-select"
+                        value={audioMonitorForGUI}
+                        onChange={(e) => {
+                            setAudioMonitorForGUI(e.target.value);
+                            setItem(INDEXEDDB_KEY_AUDIO_MONITR, e.target.value);
+                        }}
+                    >
+                        {outputAudioDeviceInfo.map((x) => {
+                            return (
+                                <option key={x.deviceId} value={x.deviceId}>
+                                    {x.label}
+                                </option>
+                            );
+                        })}
+                    </select>
+                </div>
+            </div>
+        );
+    }, [serverSetting.serverSetting.enableServerAudio, outputAudioDeviceInfo, audioMonitorForGUI]);
+
+    useEffect(() => {
+        console.log("initializedRef.current", initializedRef.current);
+        setAudioMonitorElementId(AUDIO_ELEMENT_FOR_PLAY_MONITOR);
+    }, [initializedRef.current]);
+
+    // (6-2) サーバ
     const serverMonitorRow = useMemo(() => {
         if (serverSetting.serverSetting.enableServerAudio == 0) {
             return <></>;
@@ -675,6 +757,41 @@ export const DeviceArea = (_props: DeviceAreaProps) => {
         );
     }, [monitorHostApi, serverSetting.serverSetting, serverSetting.updateServerSettings, serverSetting.serverSetting.enableServerAudio]);
 
+    const monitorGainControl = useMemo(() => {
+        const currentMonitorGain = serverSetting.serverSetting.enableServerAudio == 0 ? setting.voiceChangerClientSetting.monitorGain : serverSetting.serverSetting.serverMonitorAudioGain;
+        const monitorValueUpdatedAction =
+            serverSetting.serverSetting.enableServerAudio == 0
+                ? async (val: number) => {
+                      await setVoiceChangerClientSetting({ ...setting.voiceChangerClientSetting, monitorGain: val });
+                  }
+                : async (val: number) => {
+                      await serverSetting.updateServerSettings({ ...serverSetting.serverSetting, serverMonitorAudioGain: val });
+                  };
+
+        return (
+            <div className="config-sub-area-control">
+                <div className="config-sub-area-control-title left-padding-2">gain</div>
+                <div className="config-sub-area-control-field">
+                    <div className="config-sub-area-control-field-auido-io">
+                        <span className="character-area-slider-control-slider">
+                            <input
+                                type="range"
+                                min="0.1"
+                                max="10.0"
+                                step="0.1"
+                                value={currentMonitorGain}
+                                onChange={(e) => {
+                                    monitorValueUpdatedAction(Number(e.target.value));
+                                }}
+                            ></input>
+                        </span>
+                        <span className="character-area-slider-control-val">{currentMonitorGain}</span>
+                    </div>
+                </div>
+            </div>
+        );
+    }, [serverSetting.serverSetting, setting, setVoiceChangerClientSetting, serverSetting.updateServerSettings]);
+
     return (
         <div className="config-sub-area">
             {deviceModeRow}
@@ -685,10 +802,13 @@ export const DeviceArea = (_props: DeviceAreaProps) => {
             {audioInputScreenRow}
             {clientAudioOutputRow}
             {serverAudioOutputRow}
+            {clientMonitorRow}
             {serverMonitorRow}
+            {monitorGainControl}
 
             {outputRecorderRow}
             <audio hidden id={AUDIO_ELEMENT_FOR_PLAY_RESULT}></audio>
+            <audio hidden id={AUDIO_ELEMENT_FOR_PLAY_MONITOR}></audio>
         </div>
     );
 };

client/demo/src/const.ts

@@ -1,13 +1,15 @@
 
-export const AUDIO_ELEMENT_FOR_PLAY_RESULT = "audio-result"
+export const AUDIO_ELEMENT_FOR_PLAY_RESULT = "audio-result"  // 変換後の出力用プレイヤー
-export const AUDIO_ELEMENT_FOR_TEST_ORIGINAL = "audio-test-original"
+export const AUDIO_ELEMENT_FOR_PLAY_MONITOR = "audio-monitor"  //  変換後のモニター用プレイヤー
-export const AUDIO_ELEMENT_FOR_TEST_CONVERTED = "audio-test-converted"
+export const AUDIO_ELEMENT_FOR_TEST_ORIGINAL = "audio-test-original"  // ??? 使ってないかも。
-export const AUDIO_ELEMENT_FOR_TEST_CONVERTED_ECHOBACK = "audio-test-converted-echoback"
+export const AUDIO_ELEMENT_FOR_TEST_CONVERTED = "audio-test-converted" // ファイルインプットのコントロール
+export const AUDIO_ELEMENT_FOR_TEST_CONVERTED_ECHOBACK = "audio-test-converted-echoback"  // ファイルインプットのエコーバック
 
 export const AUDIO_ELEMENT_FOR_SAMPLING_INPUT = "body-wav-container-wav-input"
 export const AUDIO_ELEMENT_FOR_SAMPLING_OUTPUT = "body-wav-container-wav-output"
 
 export const INDEXEDDB_KEY_AUDIO_OUTPUT = "INDEXEDDB_KEY_AUDIO_OUTPUT"
+export const INDEXEDDB_KEY_AUDIO_MONITR = "INDEXEDDB_KEY_AUDIO_MONITOR"
 export const INDEXEDDB_KEY_DEFAULT_MODEL_TYPE = "INDEXEDDB_KEY_DEFALT_MODEL_TYPE"
 
 

client/demo/src/css/App.css

@@ -757,6 +757,18 @@ body {
                 max-height: 60vh;
                 width: 100%;
                 overflow-y: scroll;
+                &::-webkit-scrollbar {
+                    width: 10px;
+                    height: 10px;
+                }
+                &::-webkit-scrollbar-track {
+                    background-color: #eee;
+                    border-radius: 3px;
+                }
+                &::-webkit-scrollbar-thumb {
+                    background: #f7cfec80;
+                    border-radius: 3px;
+                }
 
                 .model-slot {
                     height: 5rem;
@@ -1150,12 +1162,30 @@ body {
             flex-direction: row;
             gap: 2px;
             flex-wrap: wrap;
+            overflow-y: scroll;
+            max-height: 12rem;
+            &::-webkit-scrollbar {
+                width: 10px;
+                height: 10px;
+            }
+            &::-webkit-scrollbar-track {
+                background-color: #eee;
+                border-radius: 3px;
+            }
+            &::-webkit-scrollbar-thumb {
+                background: #f7cfec80;
+                border-radius: 3px;
+            }
+
             /* width: calc(30rem + 40px + 10px); */
         }
 
         .model-slot-buttons {
             display: flex;
-            flex-direction: column-reverse;
+            gap: 5px;
+            flex-direction: column;
+            justify-content: space-between;
+            width: 4rem;
             .model-slot-button {
                 border: solid 2px #999;
                 color: white;
@@ -1164,10 +1194,41 @@ body {
                 background: #333;
                 cursor: pointer;
                 padding: 5px;
+                text-align: center;
+                width: 3rem;
             }
             .model-slot-button:hover {
                 border: solid 2px #faa;
             }
+            .model-slot-sort-buttons {
+                height: 50%;
+                .model-slot-sort-button {
+                    color: white;
+                    font-size: 0.8rem;
+                    border-radius: 4px;
+                    background: #333;
+                    border: solid 2px #444;
+                    cursor: pointer;
+                    padding: 1px;
+                    text-align: center;
+                    width: 3rem;
+                }
+                .model-slot-sort-button-active {
+                    color: white;
+                    font-size: 0.8rem;
+                    border-radius: 4px;
+                    background: #595;
+                    border: solid 2px #595;
+                    cursor: pointer;
+                    padding: 1px;
+                    text-align: center;
+                    width: 3rem;
+                }
+                .model-slot-sort-button:hover {
+                    border: solid 2px #faa;
+                    background: #343;
+                }
+            }
         }
     }
 }
@@ -1277,6 +1338,7 @@ body {
         .character-area-control {
             display: flex;
             gap: 3px;
+            align-items: center;
             .character-area-control-buttons {
                 display: flex;
                 flex-direction: row;
@@ -1301,6 +1363,34 @@ body {
                         border: solid 1px #000;
                     }
                 }
+                .character-area-control-passthru-button-stanby {
+                    width: 5rem;
+                    border: solid 1px #999;
+                    border-radius: 15px;
+                    padding: 2px;
+                    background: #aba;
+                    cursor: pointer;
+                    font-weight: 700;
+                    font-size: 0.8rem;
+                    text-align: center;
+                    &:hover {
+                        border: solid 1px #000;
+                    }
+                }
+                .character-area-control-passthru-button-active {
+                    width: 5rem;
+                    border: solid 1px #955;
+                    border-radius: 15px;
+                    padding: 2px;
+                    background: #fdd;
+                    cursor: pointer;
+                    font-weight: 700;
+                    font-size: 0.8rem;
+                    text-align: center;
+                    &:hover {
+                        border: solid 1px #000;
+                    }
+                }
             }
 
             .character-area-control-title {
@@ -1344,6 +1434,35 @@ body {
                     .character-area-button:hover {
                         border: solid 2px #faa;
                     }
+                    .character-area-toggle-button {
+                        border: solid 2px #999;
+                        color: white;
+                        background: #666;
+
+                        cursor: pointer;
+
+                        font-size: 0.8rem;
+                        border-radius: 5px;
+                        height: 1.2rem;
+                        padding-left: 2px;
+                        padding-right: 2px;
+                    }
+                    .character-area-toggle-button:hover {
+                        border: solid 2px #faa;
+                    }
+                    .character-area-toggle-button-active {
+                        border: solid 2px #999;
+                        color: white;
+                        background: #844;
+
+                        cursor: pointer;
+
+                        font-size: 0.8rem;
+                        border-radius: 5px;
+                        height: 1.2rem;
+                        padding-left: 2px;
+                        padding-right: 2px;
+                    }
                 }
             }
         }
@@ -1443,6 +1562,10 @@ audio::-webkit-media-controls-overlay-enclosure{
                         height: 1.2rem;
                         padding-left: 2px;
                         padding-right: 2px;
+                        white-space: nowrap;
+                    }
+                    .config-sub-area-button-text-small {
+                        font-size: 0.5rem;
                     }
                 }
                 .config-sub-area-control-field-auido-io {
@@ -1635,6 +1758,21 @@ audio::-webkit-media-controls-overlay-enclosure{
         flex-direction: row;
         .merge-lab-model-list {
             width: 70%;
+            overflow-y: scroll;
+            max-height: 20rem;
+            &::-webkit-scrollbar {
+                width: 10px;
+                height: 10px;
+            }
+            &::-webkit-scrollbar-track {
+                background-color: #eee;
+                border-radius: 3px;
+            }
+            &::-webkit-scrollbar-thumb {
+                background: #f7cfec80;
+                border-radius: 3px;
+            }
+
             .merge-lab-model-item {
                 display: flex;
                 flex-direction: row;
@@ -1673,3 +1811,18 @@ audio::-webkit-media-controls-overlay-enclosure{
         }
     }
 }
+
+.blinking {
+    animation: flash 0.7s cubic-bezier(0.91, -0.14, 0, 1.4) infinite;
+}
+
+@keyframes flash {
+    0%,
+    100% {
+        opacity: 1;
+    }
+
+    50% {
+        opacity: 0.5;
+    }
+}

client/lib/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@dannadori/voice-changer-client-js",
-  "version": "1.0.164",
+  "version": "1.0.167",
   "description": "",
   "main": "dist/index.js",
   "directories": {
@@ -26,33 +26,33 @@
   "author": "wataru.okada@flect.co.jp",
   "license": "ISC",
   "devDependencies": {
-    "@types/audioworklet": "^0.0.48",
+    "@types/audioworklet": "^0.0.50",
-    "@types/node": "^20.4.2",
+    "@types/node": "^20.4.8",
-    "@types/react": "18.2.15",
+    "@types/react": "18.2.18",
     "@types/react-dom": "18.2.7",
-    "eslint": "^8.45.0",
+    "eslint": "^8.46.0",
-    "eslint-config-prettier": "^8.8.0",
+    "eslint-config-prettier": "^9.0.0",
     "eslint-plugin-prettier": "^5.0.0",
-    "eslint-plugin-react": "^7.32.2",
+    "eslint-plugin-react": "^7.33.1",
     "eslint-webpack-plugin": "^4.0.1",
     "npm-run-all": "^4.1.5",
-    "prettier": "^3.0.0",
+    "prettier": "^3.0.1",
     "raw-loader": "^4.0.2",
     "rimraf": "^5.0.1",
     "ts-loader": "^9.4.4",
     "typescript": "^5.1.6",
-    "webpack": "^5.88.1",
+    "webpack": "^5.88.2",
     "webpack-cli": "^5.1.4",
     "webpack-dev-server": "^4.15.1"
   },
   "dependencies": {
-    "@types/readable-stream": "^2.3.15",
+    "@types/readable-stream": "^4.0.0",
     "amazon-chime-sdk-js": "^3.15.0",
     "buffer": "^6.0.3",
     "localforage": "^1.10.0",
     "protobufjs": "^7.2.4",
     "react": "^18.2.0",
     "react-dom": "^18.2.0",
-    "socket.io-client": "^4.7.1"
+    "socket.io-client": "^4.7.2"
   }
 }

client/lib/src/VoiceChangerClient.ts

@@ -23,9 +23,11 @@ export class VoiceChangerClient {
     private currentMediaStreamAudioSourceNode: MediaStreamAudioSourceNode | null = null
     private inputGainNode: GainNode | null = null
     private outputGainNode: GainNode | null = null
+    private monitorGainNode: GainNode | null = null
     private vcInNode!: VoiceChangerWorkletNode
     private vcOutNode!: VoiceChangerWorkletNode
     private currentMediaStreamAudioDestinationNode!: MediaStreamAudioDestinationNode
+    private currentMediaStreamAudioDestinationMonitorNode!: MediaStreamAudioDestinationNode
 
 
     private promiseForInitialize: Promise<void>
@@ -72,6 +74,12 @@ export class VoiceChangerClient {
             this.vcOutNode.connect(this.outputGainNode) // vc node -> output node
             this.outputGainNode.connect(this.currentMediaStreamAudioDestinationNode)
 
+            this.currentMediaStreamAudioDestinationMonitorNode = ctx44k.createMediaStreamDestination() // output node
+            this.monitorGainNode = ctx44k.createGain()
+            this.monitorGainNode.gain.value = this.setting.monitorGain
+            this.vcOutNode.connect(this.monitorGainNode) // vc node -> monitor node
+            this.monitorGainNode.connect(this.currentMediaStreamAudioDestinationMonitorNode)
+
             if (this.vfEnable) {
                 this.vf = await VoiceFocusDeviceTransformer.create({ variant: 'c20' })
                 const dummyMediaStream = createDummyMediaStream(this.ctx)
@@ -185,6 +193,9 @@ export class VoiceChangerClient {
     get stream(): MediaStream {
         return this.currentMediaStreamAudioDestinationNode.stream
     }
+    get monitorStream(): MediaStream {
+        return this.currentMediaStreamAudioDestinationMonitorNode.stream
+    }
 
     start = async () => {
         await this.vcInNode.start()
@@ -239,6 +250,9 @@ export class VoiceChangerClient {
         if (this.setting.outputGain != setting.outputGain) {
             this.setOutputGain(setting.outputGain)
         }
+        if (this.setting.monitorGain != setting.monitorGain) {
+            this.setMonitorGain(setting.monitorGain)
+        }
 
         this.setting = setting
         if (reconstructInputRequired) {
@@ -251,6 +265,9 @@ export class VoiceChangerClient {
         if (!this.inputGainNode) {
             return
         }
+        if(!val){
+            return
+        }
         this.inputGainNode.gain.value = val
     }
 
@@ -258,9 +275,22 @@ export class VoiceChangerClient {
         if (!this.outputGainNode) {
             return
         }
+        if(!val){
+            return
+        }
         this.outputGainNode.gain.value = val
     }
 
+    setMonitorGain = (val: number) => {
+        if (!this.monitorGainNode) {
+            return
+        }
+        if(!val){
+            return
+        }
+        this.monitorGainNode.gain.value = val
+    }
+
     /////////////////////////////////////////////////////
     // コンポーネント設定、操作
     /////////////////////////////////////////////////////

client/lib/src/const.ts

@@ -68,6 +68,7 @@ export const RVCModelType = {
 export type RVCModelType = typeof RVCModelType[keyof typeof RVCModelType]
 
 export const ServerSettingKey = {
+    "passThrough":"passThrough",
     "srcId": "srcId",
     "dstId": "dstId",
     "gpu": "gpu",
@@ -97,6 +98,7 @@ export const ServerSettingKey = {
     "serverReadChunkSize": "serverReadChunkSize",
     "serverInputAudioGain": "serverInputAudioGain",
     "serverOutputAudioGain": "serverOutputAudioGain",
+    "serverMonitorAudioGain": "serverMonitorAudioGain",
 
     "tran": "tran",
     "noiseScale": "noiseScale",
@@ -123,6 +125,7 @@ export const ServerSettingKey = {
     "threshold": "threshold",
 
     "speedUp": "speedUp",
+    "skipDiffusion": "skipDiffusion",
 
     "inputSampleRate": "inputSampleRate",
     "enableDirectML": "enableDirectML",
@@ -131,6 +134,7 @@ export type ServerSettingKey = typeof ServerSettingKey[keyof typeof ServerSettin
 
 
 export type VoiceChangerServerSetting = {
+    passThrough: boolean
     srcId: number,
     dstId: number,
     gpu: number,
@@ -157,6 +161,7 @@ export type VoiceChangerServerSetting = {
     serverReadChunkSize: number
     serverInputAudioGain: number
     serverOutputAudioGain: number
+    serverMonitorAudioGain: number
 
 
     tran: number // so-vits-svc
@@ -184,13 +189,14 @@ export type VoiceChangerServerSetting = {
     threshold: number// DDSP-SVC
 
     speedUp: number // Diffusion-SVC
-
+    skipDiffusion: number // Diffusion-SVC 0:off, 1:on
 
     inputSampleRate: InputSampleRate
     enableDirectML: number
 }
 
 type ModelSlot = {
+    slotIndex: number
     voiceChangerType: VoiceChangerType
     name: string,
     description: string,
@@ -303,7 +309,9 @@ export type ServerInfo = VoiceChangerServerSetting & {
         memory: number,
     }[]
     maxInputLength: number  // MMVCv15
-
+    voiceChangerParams: {
+        model_dir: string
+    }
 }
 
 export type SampleModel = {
@@ -339,6 +347,7 @@ export type DiffusionSVCSampleModel =SampleModel & {
 
 export const DefaultServerSetting: ServerInfo = {
     // VC Common 
+    passThrough: false,
     inputSampleRate: 48000,
 
     crossFadeOffsetRate: 0.0,
@@ -361,6 +370,7 @@ export const DefaultServerSetting: ServerInfo = {
     serverReadChunkSize: 256,
     serverInputAudioGain: 1.0,
     serverOutputAudioGain: 1.0,
+    serverMonitorAudioGain: 1.0,
 
     // VC Specific
     srcId: 0,
@@ -397,6 +407,7 @@ export const DefaultServerSetting: ServerInfo = {
     threshold: -45,
 
     speedUp: 10,
+    skipDiffusion: 1,
 
     enableDirectML: 0,
     // 
@@ -405,7 +416,10 @@ export const DefaultServerSetting: ServerInfo = {
     serverAudioInputDevices: [],
     serverAudioOutputDevices: [],
 
-    maxInputLength:  128 * 2048
+    maxInputLength:  128 * 2048,
+    voiceChangerParams: {
+        model_dir: ""
+    }
 }
 
 ///////////////////////
@@ -466,6 +480,7 @@ export type VoiceChangerClientSetting = {
 
     inputGain: number
     outputGain: number
+    monitorGain: number
 }
 
 ///////////////////////
@@ -496,7 +511,8 @@ export const DefaultClientSettng: ClientSetting = {
         noiseSuppression: false,
         noiseSuppression2: false,
         inputGain: 1.0,
-        outputGain: 1.0
+        outputGain: 1.0,
+        monitorGain: 1.0
     }
 }
 
@@ -533,7 +549,7 @@ export type OnnxExporterInfo = {
 
 // Merge
 export type MergeElement = {
-    filename: string
+    slotIndex: number
     strength: number
 }
 export type MergeModelRequest = {

client/lib/src/hooks/useClient.ts

@@ -47,6 +47,7 @@ export type ClientState = {
     clearSetting: () => Promise<void>
     // AudioOutputElement  設定
     setAudioOutputElementId: (elemId: string) => void
+    setAudioMonitorElementId: (elemId: string) => void
 
     ioErrorCount: number
     resetIoErrorCount: () => void
@@ -215,6 +216,18 @@ export const useClient = (props: UseClientProps): ClientState => {
         }
     }
 
+    const setAudioMonitorElementId = (elemId: string) => {
+        if (!voiceChangerClientRef.current) {
+            console.warn("[voiceChangerClient] is not ready for set audio output.")
+            return
+        }
+        const audio = document.getElementById(elemId) as HTMLAudioElement
+        if (audio.paused) {
+            audio.srcObject = voiceChangerClientRef.current.monitorStream
+            audio.play()
+        }
+    }
+
     // (2-2) 情報リロード
     const getInfo = useMemo(() => {
         return async () => {
@@ -286,6 +299,7 @@ export const useClient = (props: UseClientProps): ClientState => {
 
         // AudioOutputElement  設定
         setAudioOutputElementId,
+        setAudioMonitorElementId,
 
         ioErrorCount,
         resetIoErrorCount

docker_vcclient/README.md

@@ -18,6 +18,10 @@ npm run build:docker:vcclient
 bash start_docker.sh
 ```
 
+ブラウザ(Chrome のみサポート)でアクセスすると画面が表示されます。
+
+## RUN with options
+
 GPU を使用しない場合は
 
 ```

docker_vcclient/README_en.md

@@ -36,6 +36,10 @@ In root folder of repos.
 bash start_docker.sh
 ```
 
+Access with Browser (currently only chrome is supported), then you can see gui.
+
+## RUN with options
+
 Without GPU
 
 ```

server/MMVCServerSIO.py

@@ -9,6 +9,7 @@ import argparse
 from Exceptions import WeightDownladException
 from downloader.SampleDownloader import downloadInitialSamples
 from downloader.WeightDownloader import downloadWeight
+from voice_changer.VoiceChangerParamsManager import VoiceChangerParamsManager
 
 from voice_changer.utils.VoiceChangerParams import VoiceChangerParams
 
@@ -40,19 +41,19 @@ def setupArgParser():
     parser.add_argument("--httpsCert", type=str, default="ssl.cert", help="path for the cert of https")
     parser.add_argument("--httpsSelfSigned", type=strtobool, default=True, help="generate self-signed certificate")
 
-    parser.add_argument("--model_dir", type=str, help="path to model files")
+    parser.add_argument("--model_dir", type=str, default="model_dir", help="path to model files")
     parser.add_argument("--sample_mode", type=str, default="production", help="rvc_sample_mode")
 
-    parser.add_argument("--content_vec_500", type=str, help="path to content_vec_500 model(pytorch)")
+    parser.add_argument("--content_vec_500", type=str, default="pretrain/checkpoint_best_legacy_500.pt", help="path to content_vec_500 model(pytorch)")
-    parser.add_argument("--content_vec_500_onnx", type=str, help="path to content_vec_500 model(onnx)")
+    parser.add_argument("--content_vec_500_onnx", type=str, default="pretrain/content_vec_500.onnx", help="path to content_vec_500 model(onnx)")
-    parser.add_argument("--content_vec_500_onnx_on", type=strtobool, default=False, help="use or not onnx for  content_vec_500")
+    parser.add_argument("--content_vec_500_onnx_on", type=strtobool, default=True, help="use or not onnx for  content_vec_500")
-    parser.add_argument("--hubert_base", type=str, help="path to hubert_base model(pytorch)")
+    parser.add_argument("--hubert_base", type=str, default="pretrain/hubert_base.pt", help="path to hubert_base model(pytorch)")
-    parser.add_argument("--hubert_base_jp", type=str, help="path to hubert_base_jp model(pytorch)")
+    parser.add_argument("--hubert_base_jp", type=str, default="pretrain/rinna_hubert_base_jp.pt", help="path to hubert_base_jp model(pytorch)")
-    parser.add_argument("--hubert_soft", type=str, help="path to hubert_soft model(pytorch)")
+    parser.add_argument("--hubert_soft", type=str, default="pretrain/hubert/hubert-soft-0d54a1f4.pt", help="path to hubert_soft model(pytorch)")
-    parser.add_argument("--nsf_hifigan", type=str, help="path to nsf_hifigan model(pytorch)")
+    parser.add_argument("--nsf_hifigan", type=str, default="pretrain/nsf_hifigan/model", help="path to nsf_hifigan model(pytorch)")
-    parser.add_argument("--crepe_onnx_full", type=str, help="path to crepe_onnx_full")
+    parser.add_argument("--crepe_onnx_full", type=str, default="pretrain/crepe_onnx_full.onnx", help="path to crepe_onnx_full")
-    parser.add_argument("--crepe_onnx_tiny", type=str, help="path to crepe_onnx_tiny")
+    parser.add_argument("--crepe_onnx_tiny", type=str, default="pretrain/crepe_onnx_tiny.onnx", help="path to crepe_onnx_tiny")
-    parser.add_argument("--rmvpe", type=str, help="path to rmvpe")
+    parser.add_argument("--rmvpe", type=str, default="pretrain/rmvpe.pt", help="path to rmvpe")
 
     return parser
 
@@ -96,6 +97,8 @@ voiceChangerParams = VoiceChangerParams(
     rmvpe=args.rmvpe,
     sample_mode=args.sample_mode,
 )
+vcparams = VoiceChangerParamsManager.get_instance()
+vcparams.setParams(voiceChangerParams)
 
 printMessage(f"Booting PHASE :{__name__}", level=2)
 
@@ -124,7 +127,8 @@ if __name__ == "MMVCServerSIO":
 
 
 if __name__ == "__mp_main__":
-    printMessage("サーバプロセスを起動しています。", level=2)
+    # printMessage("サーバプロセスを起動しています。", level=2)
+    printMessage("The server process is starting up.", level=2)
 
 if __name__ == "__main__":
     mp.freeze_support()
@@ -132,12 +136,13 @@ if __name__ == "__main__":
     logger.debug(args)
 
     printMessage(f"PYTHON:{sys.version}", level=2)
-    printMessage("Voice Changerを起動しています。", level=2)
+    # printMessage("Voice Changerを起動しています。", level=2)
+    printMessage("Activating the Voice Changer.", level=2)
     # ダウンロード(Weight)
     try:
         downloadWeight(voiceChangerParams)
     except WeightDownladException:
-        printMessage("RVC用のモデルファイルのダウンロードに失敗しました。", level=2)
+        # printMessage("RVC用のモデルファイルのダウンロードに失敗しました。", level=2)
         printMessage("failed to download weight for rvc", level=2)
 
     # ダウンロード(Sample)
@@ -192,29 +197,31 @@ if __name__ == "__main__":
     printMessage("-- ---- -- ", level=1)
 
     # アドレス表示
-    printMessage("ブラウザで次のURLを開いてください.", level=2)
+    printMessage("Please open the following URL in your browser.", level=2)
+    # printMessage("ブラウザで次のURLを開いてください.", level=2)
     if args.https == 1:
         printMessage("https://<IP>:<PORT>/", level=1)
     else:
         printMessage("http://<IP>:<PORT>/", level=1)
 
-    printMessage("多くの場合は次のいずれかのURLにアクセスすると起動します。", level=2)
+    # printMessage("多くの場合は次のいずれかのURLにアクセスすると起動します。", level=2)
+    printMessage("In many cases, it will launch when you access any of the following URLs.", level=2)
     if "EX_PORT" in locals() and "EX_IP" in locals():  # シェルスクリプト経由起動(docker)
         if args.https == 1:
-            printMessage(f"https://localhost:{EX_PORT}/", level=1)
+            printMessage(f"https://127.0.0.1:{EX_PORT}/", level=1)
             for ip in EX_IP.strip().split(" "):
                 printMessage(f"https://{ip}:{EX_PORT}/", level=1)
         else:
-            printMessage(f"http://localhost:{EX_PORT}/", level=1)
+            printMessage(f"http://127.0.0.1:{EX_PORT}/", level=1)
     else:  # 直接python起動
         if args.https == 1:
             s = socket.socket(socket.AF_INET, socket.SOCK_DGRAM)
             s.connect((args.test_connect, 80))
             hostname = s.getsockname()[0]
-            printMessage(f"https://localhost:{PORT}/", level=1)
+            printMessage(f"https://127.0.0.1:{PORT}/", level=1)
             printMessage(f"https://{hostname}:{PORT}/", level=1)
         else:
-            printMessage(f"http://localhost:{PORT}/", level=1)
+            printMessage(f"http://127.0.0.1:{PORT}/", level=1)
 
     # サーバ起動
     if args.https:
@@ -237,15 +244,15 @@ if __name__ == "__main__":
         p.start()
         try:
             if sys.platform.startswith("win"):
-                process = subprocess.Popen([NATIVE_CLIENT_FILE_WIN, "--disable-gpu", "-u", f"http://localhost:{PORT}/"])
+                process = subprocess.Popen([NATIVE_CLIENT_FILE_WIN, "--disable-gpu", "-u", f"http://127.0.0.1:{PORT}/"])
                 return_code = process.wait()
-                logger.info("client closed.")                
+                logger.info("client closed.")
                 p.terminate()
             elif sys.platform.startswith("darwin"):
-                process = subprocess.Popen([NATIVE_CLIENT_FILE_MAC, "--disable-gpu", "-u", f"http://localhost:{PORT}/"])
+                process = subprocess.Popen([NATIVE_CLIENT_FILE_MAC, "--disable-gpu", "-u", f"http://127.0.0.1:{PORT}/"])
                 return_code = process.wait()
-                logger.info("client closed.")                
+                logger.info("client closed.")
                 p.terminate()
 
         except Exception as e:
-            logger.error(f"[Voice Changer] Launch Exception, {e}")
+            logger.error(f"[Voice Changer] Client Launch Exception, {e}")

server/const.py

@@ -169,4 +169,4 @@ def getSampleJsonAndModelIds(mode: RVCSampleMode):
 
 
 RVC_MODEL_DIRNAME = "rvc"
-MAX_SLOT_NUM = 10
+MAX_SLOT_NUM = 200

server/data/ModelSlot.py

@@ -9,6 +9,7 @@ import json
 
 @dataclass
 class ModelSlot:
+    slotIndex: int = -1
     voiceChangerType: VoiceChangerType | None = None
     name: str = ""
     description: str = ""
@@ -132,19 +133,26 @@ def loadSlotInfo(model_dir: str, slotIndex: int) -> ModelSlots:
     if not os.path.exists(jsonFile):
         return ModelSlot()
     jsonDict = json.load(open(os.path.join(slotDir, "params.json")))
-    slotInfo = ModelSlot(**{k: v for k, v in jsonDict.items() if k in ModelSlot.__annotations__})
+    slotInfoKey = list(ModelSlot.__annotations__.keys())
+    slotInfo = ModelSlot(**{k: v for k, v in jsonDict.items() if k in slotInfoKey})
     if slotInfo.voiceChangerType == "RVC":
-        return RVCModelSlot(**jsonDict)
+        slotInfoKey.extend(list(RVCModelSlot.__annotations__.keys()))
+        return RVCModelSlot(**{k: v for k, v in jsonDict.items() if k in slotInfoKey})
     elif slotInfo.voiceChangerType == "MMVCv13":
-        return MMVCv13ModelSlot(**jsonDict)
+        slotInfoKey.extend(list(MMVCv13ModelSlot.__annotations__.keys()))
+        return MMVCv13ModelSlot(**{k: v for k, v in jsonDict.items() if k in slotInfoKey})
     elif slotInfo.voiceChangerType == "MMVCv15":
-        return MMVCv15ModelSlot(**jsonDict)
+        slotInfoKey.extend(list(MMVCv15ModelSlot.__annotations__.keys()))
+        return MMVCv15ModelSlot(**{k: v for k, v in jsonDict.items() if k in slotInfoKey})
     elif slotInfo.voiceChangerType == "so-vits-svc-40":
-        return SoVitsSvc40ModelSlot(**jsonDict)
+        slotInfoKey.extend(list(SoVitsSvc40ModelSlot.__annotations__.keys()))
+        return SoVitsSvc40ModelSlot(**{k: v for k, v in jsonDict.items() if k in slotInfoKey})
     elif slotInfo.voiceChangerType == "DDSP-SVC":
-        return DDSPSVCModelSlot(**jsonDict)
+        slotInfoKey.extend(list(DDSPSVCModelSlot.__annotations__.keys()))
+        return DDSPSVCModelSlot(**{k: v for k, v in jsonDict.items() if k in slotInfoKey})
     elif slotInfo.voiceChangerType == "Diffusion-SVC":
-        return DiffusionSVCModelSlot(**jsonDict)
+        slotInfoKey.extend(list(DiffusionSVCModelSlot.__annotations__.keys()))
+        return DiffusionSVCModelSlot(**{k: v for k, v in jsonDict.items() if k in slotInfoKey})
     else:
         return ModelSlot()
 
@@ -153,10 +161,13 @@ def loadAllSlotInfo(model_dir: str):
     slotInfos: list[ModelSlots] = []
     for slotIndex in range(MAX_SLOT_NUM):
         slotInfo = loadSlotInfo(model_dir, slotIndex)
+        slotInfo.slotIndex = slotIndex  # スロットインデックスは動的に注入
         slotInfos.append(slotInfo)
     return slotInfos
 
 
 def saveSlotInfo(model_dir: str, slotIndex: int, slotInfo: ModelSlots):
     slotDir = os.path.join(model_dir, str(slotIndex))
-    json.dump(asdict(slotInfo), open(os.path.join(slotDir, "params.json"), "w"))
+    slotInfoDict = asdict(slotInfo)
+    slotInfo.slotIndex = -1  # スロットインデックスは動的に注入
+    json.dump(slotInfoDict, open(os.path.join(slotDir, "params.json"), "w"), indent=4)

server/downloader/SampleDownloader.py

@@ -1,5 +1,6 @@
 import json
 import os
+import sys
 from concurrent.futures import ThreadPoolExecutor
 from typing import Any, Tuple
 
@@ -7,7 +8,6 @@ from const import RVCSampleMode, getSampleJsonAndModelIds
 from data.ModelSample import ModelSamples, generateModelSample
 from data.ModelSlot import DiffusionSVCModelSlot, ModelSlot, RVCModelSlot
 from mods.log_control import VoiceChangaerLogger
-from voice_changer.DiffusionSVC.DiffusionSVCModelSlotGenerator import DiffusionSVCModelSlotGenerator
 from voice_changer.ModelSlotManager import ModelSlotManager
 from voice_changer.RVC.RVCModelSlotGenerator import RVCModelSlotGenerator
 from downloader.Downloader import download, download_no_tqdm
@@ -109,7 +109,7 @@ def _downloadSamples(samples: list[ModelSamples], sampleModelIds: list[Tuple[str
                     "position": line_num,
                 }
             )
-            slotInfo.modelFile = modelFilePath
+            slotInfo.modelFile = os.path.basename(sample.modelUrl)
             line_num += 1
 
             if targetSampleParams["useIndex"] is True and hasattr(sample, "indexUrl") and sample.indexUrl != "":
@@ -124,7 +124,7 @@ def _downloadSamples(samples: list[ModelSamples], sampleModelIds: list[Tuple[str
                         "position": line_num,
                     }
                 )
-                slotInfo.indexFile = indexPath
+                slotInfo.indexFile = os.path.basename(sample.indexUrl)
                 line_num += 1
 
             if hasattr(sample, "icon") and sample.icon != "":
@@ -139,7 +139,7 @@ def _downloadSamples(samples: list[ModelSamples], sampleModelIds: list[Tuple[str
                         "position": line_num,
                     }
                 )
-                slotInfo.iconFile = iconPath
+                slotInfo.iconFile = os.path.basename(sample.icon)
                 line_num += 1
 
             slotInfo.sampleId = sample.id
@@ -153,6 +153,8 @@ def _downloadSamples(samples: list[ModelSamples], sampleModelIds: list[Tuple[str
             slotInfo.isONNX = slotInfo.modelFile.endswith(".onnx")
             modelSlotManager.save_model_slot(targetSlotIndex, slotInfo)
         elif sample.voiceChangerType == "Diffusion-SVC":
+            if sys.platform.startswith("darwin") is True:
+                continue
             slotInfo: DiffusionSVCModelSlot = DiffusionSVCModelSlot()
 
             os.makedirs(slotDir, exist_ok=True)
@@ -167,7 +169,7 @@ def _downloadSamples(samples: list[ModelSamples], sampleModelIds: list[Tuple[str
                     "position": line_num,
                 }
             )
-            slotInfo.modelFile = modelFilePath
+            slotInfo.modelFile = os.path.basename(sample.modelUrl)
             line_num += 1
 
             if hasattr(sample, "icon") and sample.icon != "":
@@ -182,7 +184,7 @@ def _downloadSamples(samples: list[ModelSamples], sampleModelIds: list[Tuple[str
                         "position": line_num,
                     }
                 )
-                slotInfo.iconFile = iconPath
+                slotInfo.iconFile = os.path.basename(sample.icon)
                 line_num += 1
 
             slotInfo.sampleId = sample.id
@@ -212,16 +214,19 @@ def _downloadSamples(samples: list[ModelSamples], sampleModelIds: list[Tuple[str
     logger.info("[Voice Changer] Generating metadata...")
     for targetSlotIndex in slotIndex:
         slotInfo = modelSlotManager.get_slot_info(targetSlotIndex)
+        modelPath = os.path.join(model_dir, str(slotInfo.slotIndex), os.path.basename(slotInfo.modelFile))
         if slotInfo.voiceChangerType == "RVC":
             if slotInfo.isONNX:
-                slotInfo = RVCModelSlotGenerator._setInfoByONNX(slotInfo)
+                slotInfo = RVCModelSlotGenerator._setInfoByONNX(modelPath, slotInfo)
             else:
-                slotInfo = RVCModelSlotGenerator._setInfoByPytorch(slotInfo)
+                slotInfo = RVCModelSlotGenerator._setInfoByPytorch(modelPath, slotInfo)
 
             modelSlotManager.save_model_slot(targetSlotIndex, slotInfo)
         elif slotInfo.voiceChangerType == "Diffusion-SVC":
-            if slotInfo.isONNX:
+            if sys.platform.startswith("darwin") is False:
-                pass
+                from voice_changer.DiffusionSVC.DiffusionSVCModelSlotGenerator import DiffusionSVCModelSlotGenerator
-            else:
+                if slotInfo.isONNX:
-                slotInfo = DiffusionSVCModelSlotGenerator._setInfoByPytorch(slotInfo)
+                    pass
-            modelSlotManager.save_model_slot(targetSlotIndex, slotInfo)
+                else:
+                    slotInfo = DiffusionSVCModelSlotGenerator._setInfoByPytorch(slotInfo)
+                modelSlotManager.save_model_slot(targetSlotIndex, slotInfo)

server/fillSlot.sh

@@ -0,0 +1,3 @@
+for i in {1..199}; do
+    cp -r model_dir/0 model_dir/$i
+done

server/restapi/MMVC_Rest_Fileuploader.py

@@ -113,6 +113,8 @@ class MMVC_Rest_Fileuploader:
             return JSONResponse(content=json_compatible_item_data)
         except Exception as e:
             print("[Voice Changer] post_merge_models ex:", e)
+            import traceback
+            traceback.print_exc()
 
     def post_update_model_default(self):
         try:

server/voice_changer/DDSP_SVC/DDSP_SVC.py

@@ -6,6 +6,7 @@ import torch
 from data.ModelSlot import DDSPSVCModelSlot
 
 from voice_changer.DDSP_SVC.deviceManager.DeviceManager import DeviceManager
+from voice_changer.VoiceChangerParamsManager import VoiceChangerParamsManager
 
 if sys.platform.startswith("darwin"):
     baseDir = [x for x in sys.path if x.endswith("Contents/MacOS")]
@@ -69,12 +70,15 @@ class DDSP_SVC:
 
     def initialize(self):
         self.device = self.deviceManager.getDevice(self.settings.gpu)
+        vcparams = VoiceChangerParamsManager.get_instance().params
+        modelPath = os.path.join(vcparams.model_dir, str(self.slotInfo.slotIndex), "model",  self.slotInfo.modelFile)
+        diffPath = os.path.join(vcparams.model_dir, str(self.slotInfo.slotIndex), "diff", self.slotInfo.diffModelFile)
 
         self.svc_model = SvcDDSP()
         self.svc_model.setVCParams(self.params)
-        self.svc_model.update_model(self.slotInfo.modelFile, self.device)
+        self.svc_model.update_model(modelPath, self.device)
         self.diff_model = DiffGtMel(device=self.device)
-        self.diff_model.flush_model(self.slotInfo.diffModelFile, ddsp_config=self.svc_model.args)
+        self.diff_model.flush_model(diffPath, ddsp_config=self.svc_model.args)
 
     def update_settings(self, key: str, val: int | float | str):
         if key in self.settings.intData:
@@ -174,5 +178,9 @@ class DDSP_SVC:
                 if file_path.find("DDSP-SVC" + os.path.sep) >= 0:
                     # print("remove", key, file_path)
                     sys.modules.pop(key)
-            except:  # type:ignore
+            except:  # type:ignore # noqa
                 pass
+
+    def get_model_current(self):
+        return [
+        ]

server/voice_changer/DiffusionSVC/DiffusionSVC.py

@@ -14,7 +14,7 @@ from voice_changer.RVC.embedder.EmbedderManager import EmbedderManager
 # from voice_changer.RVC.onnxExporter.export2onnx import export2onnx
 from voice_changer.RVC.deviceManager.DeviceManager import DeviceManager
 
-from Exceptions import DeviceCannotSupportHalfPrecisionException, PipelineCreateException
+from Exceptions import DeviceCannotSupportHalfPrecisionException, PipelineCreateException, PipelineNotInitializedException
 
 logger = VoiceChangaerLogger.get_instance().getLogger()
 
@@ -28,7 +28,6 @@ class DiffusionSVC(VoiceChangerModel):
         InferencerManager.initialize(params)
         self.settings = DiffusionSVCSettings()
         self.params = params
-        self.pitchExtractor = PitchExtractorManager.getPitchExtractor(self.settings.f0Detector, self.settings.gpu)
 
         self.pipeline: Pipeline | None = None
 
@@ -84,6 +83,8 @@ class DiffusionSVC(VoiceChangerModel):
         if self.pipeline is not None:
             pipelineInfo = self.pipeline.getPipelineInfo()
             data["pipelineInfo"] = pipelineInfo
+        else:
+            data["pipelineInfo"] = "None"
         return data
 
     def get_processing_sampling_rate(self):
@@ -137,6 +138,9 @@ class DiffusionSVC(VoiceChangerModel):
         return (self.audio_buffer, self.pitchf_buffer, self.feature_buffer, convertSize, vol)
 
     def inference(self, receivedData: AudioInOut, crossfade_frame: int, sola_search_frame: int):
+        if self.pipeline is None:
+            logger.info("[Voice Changer] Pipeline is not initialized.")
+            raise PipelineNotInitializedException()
         data = self.generate_input(receivedData, crossfade_frame, sola_search_frame)
         audio: AudioInOut = data[0]
         pitchf: PitchfInOut = data[1]
@@ -176,7 +180,8 @@ class DiffusionSVC(VoiceChangerModel):
                 silenceFrontSec,
                 embOutputLayer,
                 useFinalProj,
-                protect
+                protect,
+                skip_diffusion=self.settings.skipDiffusion,
             )
             result = audio_out.detach().cpu().numpy()
             return result
@@ -211,6 +216,10 @@ class DiffusionSVC(VoiceChangerModel):
                 "key": "defaultTune",
                 "val": self.settings.tran,
             },
+            {
+                "key": "dstId",
+                "val": self.settings.dstId,
+            },
             {
                 "key": "defaultKstep",
                 "val": self.settings.kStep,

server/voice_changer/DiffusionSVC/DiffusionSVCModelSlotGenerator.py

@@ -1,14 +1,14 @@
 import os
-from const import EnumInferenceTypes
 from dataclasses import asdict
-import onnxruntime
+
-import json
 
 from data.ModelSlot import DiffusionSVCModelSlot, ModelSlot, RVCModelSlot
 from voice_changer.DiffusionSVC.inferencer.diffusion_svc_model.diffusion.unit2mel import load_model_vocoder_from_combo
+from voice_changer.VoiceChangerParamsManager import VoiceChangerParamsManager
 from voice_changer.utils.LoadModelParams import LoadModelParams
 from voice_changer.utils.ModelSlotGenerator import ModelSlotGenerator
 
+
 def get_divisors(n):
     divisors = []
     for i in range(1, int(n**0.5)+1):
@@ -31,6 +31,7 @@ class DiffusionSVCModelSlotGenerator(ModelSlotGenerator):
         slotInfo.name = os.path.splitext(os.path.basename(slotInfo.modelFile))[0]
         # slotInfo.iconFile = "/assets/icons/noimage.png"
         slotInfo.embChannels = 768
+        slotInfo.slotIndex = props.slot
 
         if slotInfo.isONNX:
             slotInfo = cls._setInfoByONNX(slotInfo)
@@ -40,7 +41,10 @@ class DiffusionSVCModelSlotGenerator(ModelSlotGenerator):
 
     @classmethod
     def _setInfoByPytorch(cls, slot: DiffusionSVCModelSlot):
-        diff_model, diff_args, naive_model, naive_args = load_model_vocoder_from_combo(slot.modelFile, device="cpu")
+        vcparams = VoiceChangerParamsManager.get_instance().params
+        modelPath = os.path.join(vcparams.model_dir, str(slot.slotIndex), os.path.basename(slot.modelFile))
+
+        diff_model, diff_args, naive_model, naive_args = load_model_vocoder_from_combo(modelPath, device="cpu")
         slot.kStepMax = diff_args.model.k_step_max
         slot.nLayers = diff_args.model.n_layers
         slot.nnLayers = naive_args.model.n_layers
@@ -52,53 +56,4 @@ class DiffusionSVCModelSlotGenerator(ModelSlotGenerator):
 
     @classmethod
     def _setInfoByONNX(cls, slot: ModelSlot):
-        tmp_onnx_session = onnxruntime.InferenceSession(slot.modelFile, providers=["CPUExecutionProvider"])
-        modelmeta = tmp_onnx_session.get_modelmeta()
-        try:
-            slot = RVCModelSlot(**asdict(slot))
-            metadata = json.loads(modelmeta.custom_metadata_map["metadata"])
-
-            # slot.modelType = metadata["modelType"]
-            slot.embChannels = metadata["embChannels"]
-
-            slot.embOutputLayer = metadata["embOutputLayer"] if "embOutputLayer" in metadata else 9
-            slot.useFinalProj = metadata["useFinalProj"] if "useFinalProj" in metadata else True if slot.embChannels == 256 else False
-
-            if slot.embChannels == 256:
-                slot.useFinalProj = True
-            else:
-                slot.useFinalProj = False
-
-            # ONNXモデルの情報を表示
-            if slot.embChannels == 256 and slot.embOutputLayer == 9 and slot.useFinalProj is True:
-                print("[Voice Changer] ONNX Model: Official v1 like")
-            elif slot.embChannels == 768 and slot.embOutputLayer == 12 and slot.useFinalProj is False:
-                print("[Voice Changer] ONNX Model: Official v2 like")
-            else:
-                print(f"[Voice Changer] ONNX Model: ch:{slot.embChannels}, L:{slot.embOutputLayer}, FP:{slot.useFinalProj}")
-
-            if "embedder" not in metadata:
-                slot.embedder = "hubert_base"
-            else:
-                slot.embedder = metadata["embedder"]
-
-            slot.f0 = metadata["f0"]
-            slot.modelType = EnumInferenceTypes.onnxRVC.value if slot.f0 else EnumInferenceTypes.onnxRVCNono.value
-            slot.samplingRate = metadata["samplingRate"]
-            slot.deprecated = False
-
-        except Exception as e:
-            slot.modelType = EnumInferenceTypes.onnxRVC.value
-            slot.embChannels = 256
-            slot.embedder = "hubert_base"
-            slot.f0 = True
-            slot.samplingRate = 48000
-            slot.deprecated = True
-
-            print("[Voice Changer] setInfoByONNX", e)
-            print("[Voice Changer] ############## !!!! CAUTION !!!! ####################")
-            print("[Voice Changer] This onnxfie is depricated. Please regenerate onnxfile.")
-            print("[Voice Changer] ############## !!!! CAUTION !!!! ####################")
-
-        del tmp_onnx_session
         return slot

server/voice_changer/DiffusionSVC/DiffusionSVCSettings.py

@@ -13,6 +13,7 @@ class DiffusionSVCSettings:
 
     kStep: int = 20
     speedUp: int = 10
+    skipDiffusion: int = 1  # 0:off, 1:on
 
     silenceFront: int = 1  # 0:off, 1:on
     modelSamplingRate: int = 44100
@@ -29,6 +30,7 @@ class DiffusionSVCSettings:
         "kStep",
         "speedUp",
         "silenceFront",
+        "skipDiffusion",
     ]
     floatData = ["silentThreshold"]
     strData = ["f0Detector"]

server/voice_changer/DiffusionSVC/inferencer/DiffusionSVCInferencer.py

@@ -112,25 +112,27 @@ class DiffusionSVCInferencer(Inferencer):
         k_step: int,
         infer_speedup: int,
         silence_front: float,
+        skip_diffusion: bool = True,
     ) -> torch.Tensor:
-        with Timer("pre-process") as t:
+        with Timer("pre-process", False) as t:
             gt_spec = self.naive_model_call(feats, pitch, volume, spk_id=sid, spk_mix_dict=None, aug_shift=0, spk_emb=None)
-            # gt_spec = self.vocoder.extract(audio_t, 16000)
-            # gt_spec = torch.cat((gt_spec, gt_spec[:, -1:, :]), 1)
 
         # print("[    ----Timer::1: ]", t.secs)
 
-        with Timer("pre-process") as t:
+        with Timer("pre-process", False) as t:
-            out_mel = self.__call__(feats, pitch, volume, spk_id=sid, spk_mix_dict=None, aug_shift=0, gt_spec=gt_spec, infer_speedup=infer_speedup, method='dpm-solver', k_step=k_step, use_tqdm=False, spk_emb=None)
+            if skip_diffusion == 0:
-
+                out_mel = self.__call__(feats, pitch, volume, spk_id=sid, spk_mix_dict=None, aug_shift=0, gt_spec=gt_spec, infer_speedup=infer_speedup, method='dpm-solver', k_step=k_step, use_tqdm=False, spk_emb=None)
+                gt_spec = out_mel
         # print("[    ----Timer::2: ]", t.secs)
-        with Timer("pre-process") as t:  # NOQA
+
+
+        with Timer("pre-process", False) as t:  # NOQA
             if self.vocoder_onnx is None:
                 start_frame = int(silence_front * self.vocoder.vocoder_sample_rate / self.vocoder.vocoder_hop_size)
-                out_wav = self.mel2wav(out_mel, pitch, start_frame=start_frame)
+                out_wav = self.mel2wav(gt_spec, pitch, start_frame=start_frame)
                 out_wav *= mask
             else:
-                out_wav = self.vocoder_onnx.infer(out_mel, pitch, silence_front, mask)
+                out_wav = self.vocoder_onnx.infer(gt_spec, pitch, silence_front, mask)
         # print("[    ----Timer::3: ]", t.secs)
 
         return out_wav.squeeze()

server/voice_changer/DiffusionSVC/inferencer/Inferencer.py

@@ -21,11 +21,16 @@ class Inferencer(Protocol):
 
     def infer(
         self,
+        audio_t: torch.Tensor,
         feats: torch.Tensor,
-        pitch_length: torch.Tensor,
+        pitch: torch.Tensor,
-        pitch: torch.Tensor | None,
+        volume: torch.Tensor,
-        pitchf: torch.Tensor | None,
+        mask: torch.Tensor,
         sid: torch.Tensor,
+        k_step: int,
+        infer_speedup: int,
+        silence_front: float,
+        skip_diffusion: bool = True,
     ) -> torch.Tensor:
         ...
 

server/voice_changer/DiffusionSVC/pipeline/Pipeline.py

@@ -81,23 +81,6 @@ class Pipeline(object):
 
     @torch.no_grad()
     def extract_volume_and_mask(self, audio: torch.Tensor, threshold: float):
-        '''
-        with Timer("[VolumeExt np]") as t:
-            for i in range(100):
-                volume = self.volumeExtractor.extract(audio)
-        time_np = t.secs
-        with Timer("[VolumeExt pt]") as t:
-            for i in range(100):
-                volume_t = self.volumeExtractor.extract_t(audio)
-        time_pt = t.secs
-
-        print("[Volume np]:", volume)
-        print("[Volume pt]:", volume_t)
-        print("[Perform]:", time_np, time_pt)
-        # -> [Perform]: 0.030178070068359375 0.005780220031738281 (RTX4090)
-        # -> [Perform]: 0.029046058654785156 0.0025115013122558594 (CPU i9 13900KF)
-        # ---> これくらいの処理ならCPU上のTorchでやった方が早い？
-        '''
         volume_t = self.volumeExtractor.extract_t(audio)
         mask = self.volumeExtractor.get_mask_from_volume_t(volume_t, self.inferencer_block_size, threshold=threshold)
         volume = volume_t.unsqueeze(-1).unsqueeze(0)
@@ -116,10 +99,11 @@ class Pipeline(object):
         silence_front,
         embOutputLayer,
         useFinalProj,
-        protect=0.5
+        protect=0.5,
+        skip_diffusion=True,
     ):
         # print("---------- pipe line --------------------")
-        with Timer("pre-process") as t:
+        with Timer("pre-process", False) as t:
             audio_t = torch.from_numpy(audio).float().unsqueeze(0).to(self.device)
             audio16k = self.resamplerIn(audio_t)
             volume, mask = self.extract_volume_and_mask(audio16k, threshold=-60.0)
@@ -127,7 +111,7 @@ class Pipeline(object):
             n_frames = int(audio16k.size(-1) // self.hop_size + 1)
         # print("[Timer::1: ]", t.secs)
 
-        with Timer("pre-process") as t:
+        with Timer("pre-process", False) as t:
             # ピッチ検出
             try:
                 # pitch = self.pitchExtractor.extract(
@@ -157,7 +141,7 @@ class Pipeline(object):
             feats = feats.view(1, -1)
         # print("[Timer::2: ]", t.secs)
 
-        with Timer("pre-process") as t:
+        with Timer("pre-process", False) as t:
 
             # embedding
             with autocast(enabled=self.isHalf):
@@ -175,7 +159,7 @@ class Pipeline(object):
             feats = F.interpolate(feats.permute(0, 2, 1), size=int(n_frames), mode='nearest').permute(0, 2, 1)
         # print("[Timer::3: ]", t.secs)
 
-        with Timer("pre-process") as t:
+        with Timer("pre-process", False) as t:
             # 推論実行
             try:
                 with torch.no_grad():
@@ -191,7 +175,8 @@ class Pipeline(object):
                                     sid,
                                     k_step,
                                     infer_speedup,
-                                    silence_front=silence_front
+                                    silence_front=silence_front,
+                                    skip_diffusion=skip_diffusion
                                     ).to(dtype=torch.float32),
                                 -1.0,
                                 1.0,
@@ -206,7 +191,7 @@ class Pipeline(object):
                     raise e
         # print("[Timer::4: ]", t.secs)
 
-        with Timer("pre-process") as t:  # NOQA
+        with Timer("pre-process", False) as t:  # NOQA
             feats_buffer = feats.squeeze(0).detach().cpu()
             if pitch is not None:
                 pitch_buffer = pitch.squeeze(0).detach().cpu()

server/voice_changer/DiffusionSVC/pipeline/PipelineGenerator.py

@@ -7,19 +7,23 @@ from voice_changer.DiffusionSVC.pitchExtractor.PitchExtractorManager import Pitc
 
 from voice_changer.RVC.deviceManager.DeviceManager import DeviceManager
 from voice_changer.RVC.embedder.EmbedderManager import EmbedderManager
-
+import os
 import torch
 from torchaudio.transforms import Resample
 
+from voice_changer.VoiceChangerParamsManager import VoiceChangerParamsManager
+
 
 def createPipeline(modelSlot: DiffusionSVCModelSlot, gpu: int, f0Detector: str, inputSampleRate: int, outputSampleRate: int):
     dev = DeviceManager.get_instance().getDevice(gpu)
+    vcparams = VoiceChangerParamsManager.get_instance().params
     # half = DeviceManager.get_instance().halfPrecisionAvailable(gpu)
     half = False
 
     # Inferencer 生成
-    try:
+    try:        
-        inferencer = InferencerManager.getInferencer(modelSlot.modelType, modelSlot.modelFile, gpu)
+        modelPath = os.path.join(vcparams.model_dir, str(modelSlot.slotIndex), os.path.basename(modelSlot.modelFile))
+        inferencer = InferencerManager.getInferencer(modelSlot.modelType, modelPath, gpu)
     except Exception as e:
         print("[Voice Changer] exception! loading inferencer", e)
         traceback.print_exc()

server/voice_changer/Local/ServerDevice.py

@@ -20,6 +20,13 @@ AudioDeviceKind: TypeAlias = Literal["input", "output"]
 
 logger = VoiceChangaerLogger.get_instance().getLogger()
 
+# See https://github.com/w-okada/voice-changer/issues/620
+LocalServerDeviceMode: TypeAlias = Literal[
+    "NoMonitorSeparate",
+    "WithMonitorStandard",
+    "WithMonitorAllSeparate",
+]
+
 
 @dataclass
 class ServerDeviceSettings:
@@ -39,6 +46,7 @@ class ServerDeviceSettings:
     serverReadChunkSize: int = 256
     serverInputAudioGain: float = 1.0
     serverOutputAudioGain: float = 1.0
+    serverMonitorAudioGain: float = 1.0
 
     exclusiveMode: bool = False
 
@@ -59,6 +67,7 @@ EditableServerDeviceSettings = {
     "floatData": [
         "serverInputAudioGain",
         "serverOutputAudioGain",
+        "serverMonitorAudioGain",
     ],
     "boolData": [
         "exclusiveMode"
@@ -95,6 +104,14 @@ class ServerDevice:
         self.monQueue = Queue()
         self.performance = []
 
+        # setting change確認用
+        self.currentServerInputDeviceId = -1
+        self.currentServerOutputDeviceId = -1
+        self.currentServerMonitorDeviceId = -1
+        self.currentModelSamplingRate = -1
+        self.currentInputChunkNum = -1
+        self.currentAudioSampleRate = -1
+
     def getServerInputAudioDevice(self, index: int):
         audioinput, _audiooutput = list_audio_device()
         serverAudioDevice = [x for x in audioinput if x.index == index]
@@ -111,36 +128,51 @@ class ServerDevice:
         else:
             return None
 
-    def audio_callback(self, indata: np.ndarray, outdata: np.ndarray, frames, times, status):
+    ###########################################
+    # Callback Section
+    ###########################################
+
+    def _processData(self, indata: np.ndarray):
+        indata = indata * self.settings.serverInputAudioGain
+        unpackedData = librosa.to_mono(indata.T) * 32768.0
+        unpackedData = unpackedData.astype(np.int16)
+        out_wav, times = self.serverDeviceCallbacks.on_request(unpackedData)
+        return out_wav, times
+
+    def _processDataWithTime(self, indata: np.ndarray):
+        with Timer("all_inference_time") as t:
+            out_wav, times = self._processData(indata)
+        all_inference_time = t.secs
+        self.performance = [all_inference_time] + times
+        self.serverDeviceCallbacks.emitTo(self.performance)
+        self.performance = [round(x * 1000) for x in self.performance]
+        return out_wav
+
+    def audio_callback_outQueue(self, indata: np.ndarray, outdata: np.ndarray, frames, times, status):
         try:
-            indata = indata * self.settings.serverInputAudioGain
+            out_wav = self._processDataWithTime(indata)
-            with Timer("all_inference_time") as t:
+
-                unpackedData = librosa.to_mono(indata.T) * 32768.0
+            self.outQueue.put(out_wav)
-                unpackedData = unpackedData.astype(np.int16)
+            outputChannels = outdata.shape[1]  # Monitorへのアウトプット
-                out_wav, times = self.serverDeviceCallbacks.on_request(unpackedData)
+            outdata[:] = np.repeat(out_wav, outputChannels).reshape(-1, outputChannels) / 32768.0
-                outputChannels = outdata.shape[1]
+            outdata[:] = outdata * self.settings.serverMonitorAudioGain
-                outdata[:] = np.repeat(out_wav, outputChannels).reshape(-1, outputChannels) / 32768.0
-                outdata[:] = outdata * self.settings.serverOutputAudioGain
-            all_inference_time = t.secs
-            self.performance = [all_inference_time] + times
-            self.serverDeviceCallbacks.emitTo(self.performance)
-            self.performance = [round(x * 1000) for x in self.performance]
         except Exception as e:
             print("[Voice Changer] ex:", e)
 
-    def audioInput_callback(self, indata: np.ndarray, frames, times, status):
+    def audioInput_callback_outQueue(self, indata: np.ndarray, frames, times, status):
         try:
-            indata = indata * self.settings.serverInputAudioGain
+            out_wav = self._processDataWithTime(indata)
-            with Timer("all_inference_time") as t:
+            self.outQueue.put(out_wav)
-                unpackedData = librosa.to_mono(indata.T) * 32768.0
+        except Exception as e:
-                unpackedData = unpackedData.astype(np.int16)
+            print("[Voice Changer][ServerDevice][audioInput_callback] ex:", e)
-                out_wav, times = self.serverDeviceCallbacks.on_request(unpackedData)
+            # import traceback
-                self.outQueue.put(out_wav)
+            # traceback.print_exc()
-                self.monQueue.put(out_wav)
+
-            all_inference_time = t.secs
+    def audioInput_callback_outQueue_monQueue(self, indata: np.ndarray, frames, times, status):
-            self.performance = [all_inference_time] + times
+        try:
-            self.serverDeviceCallbacks.emitTo(self.performance)
+            out_wav = self._processDataWithTime(indata)
-            self.performance = [round(x * 1000) for x in self.performance]
+            self.outQueue.put(out_wav)
+            self.monQueue.put(out_wav)
         except Exception as e:
             print("[Voice Changer][ServerDevice][audioInput_callback] ex:", e)
             # import traceback
@@ -166,15 +198,144 @@ class ServerDevice:
                 self.monQueue.get()
             outputChannels = outdata.shape[1]
             outdata[:] = np.repeat(mon_wav, outputChannels).reshape(-1, outputChannels) / 32768.0
-            outdata[:] = outdata * self.settings.serverOutputAudioGain  # GainはOutputのものをを流用
+            outdata[:] = outdata * self.settings.serverMonitorAudioGain
-            # Monitorモードが有効の場合はサンプリングレートはmonitorデバイスが優先されているためリサンプリング不要
         except Exception as e:
             print("[Voice Changer][ServerDevice][audioMonitor_callback]  ex:", e)
             # import traceback
             # traceback.print_exc()
 
+    ###########################################
+    # Main Loop Section
+    ###########################################
+    def checkSettingChanged(self):
+        if self.settings.serverAudioStated != 1:
+            print(f"serverAudioStarted Changed: {self.settings.serverAudioStated}")
+            return True
+        elif self.currentServerInputDeviceId != self.settings.serverInputDeviceId:
+            print(f"serverInputDeviceId Changed: {self.currentServerInputDeviceId} -> {self.settings.serverInputDeviceId}")
+            return True
+        elif self.currentServerOutputDeviceId != self.settings.serverOutputDeviceId:
+            print(f"serverOutputDeviceId Changed: {self.currentServerOutputDeviceId} -> {self.settings.serverOutputDeviceId}")
+            return True
+        elif self.currentServerMonitorDeviceId != self.settings.serverMonitorDeviceId:
+            print(f"serverMonitorDeviceId Changed: {self.currentServerMonitorDeviceId} -> {self.settings.serverMonitorDeviceId}")
+            return True
+        elif self.currentModelSamplingRate != self.serverDeviceCallbacks.get_processing_sampling_rate():
+            print(f"currentModelSamplingRate Changed: {self.currentModelSamplingRate} -> {self.serverDeviceCallbacks.get_processing_sampling_rate()}")
+            return True
+        elif self.currentInputChunkNum != self.settings.serverReadChunkSize:
+            print(f"currentInputChunkNum Changed: {self.currentInputChunkNum} -> {self.settings.serverReadChunkSize}")
+            return True
+        elif self.currentAudioSampleRate != self.settings.serverAudioSampleRate:
+            print(f"currentAudioSampleRate Changed: {self.currentAudioSampleRate} -> {self.settings.serverAudioSampleRate}")
+            return True
+        else:
+            return False
+
+    def runNoMonitorSeparate(self, block_frame: int, inputMaxChannel: int, outputMaxChannel: int, inputExtraSetting, outputExtraSetting):
+        with sd.InputStream(
+            callback=self.audioInput_callback_outQueue,
+            dtype="float32",
+            device=self.settings.serverInputDeviceId,
+            blocksize=block_frame,
+            samplerate=self.settings.serverInputAudioSampleRate,
+            channels=inputMaxChannel,
+            extra_settings=inputExtraSetting
+        ):
+            with sd.OutputStream(
+                callback=self.audioOutput_callback,
+                dtype="float32",
+                device=self.settings.serverOutputDeviceId,
+                blocksize=block_frame,
+                samplerate=self.settings.serverOutputAudioSampleRate,
+                channels=outputMaxChannel,
+                extra_settings=outputExtraSetting
+            ):
+                while True:
+                    changed = self.checkSettingChanged()
+                    if changed:
+                        break
+                    time.sleep(2)
+                    print(f"[Voice Changer] server audio performance {self.performance}")
+                    print(f"                status: started:{self.settings.serverAudioStated}, model_sr:{self.currentModelSamplingRate}, chunk:{self.currentInputChunkNum}")
+                    print(f"                input  : id:{self.settings.serverInputDeviceId}, sr:{self.settings.serverInputAudioSampleRate}, ch:{inputMaxChannel}")
+                    print(f"                output : id:{self.settings.serverOutputDeviceId}, sr:{self.settings.serverOutputAudioSampleRate}, ch:{outputMaxChannel}")
+                    # print(f"                monitor: id:{self.settings.serverMonitorDeviceId}, sr:{self.settings.serverMonitorAudioSampleRate}, ch:{self.serverMonitorAudioDevice.maxOutputChannels}")
+
+    def runWithMonitorStandard(self, block_frame: int, inputMaxChannel: int, outputMaxChannel: int, monitorMaxChannel: int, inputExtraSetting, outputExtraSetting, monitorExtraSetting):
+        with sd.Stream(
+            callback=self.audio_callback_outQueue,
+            dtype="float32",
+            device=(self.settings.serverInputDeviceId, self.settings.serverMonitorDeviceId),
+            blocksize=block_frame,
+            samplerate=self.settings.serverInputAudioSampleRate,
+            channels=(inputMaxChannel, monitorMaxChannel),
+            extra_settings=[inputExtraSetting, monitorExtraSetting]
+        ):
+            with sd.OutputStream(
+                callback=self.audioOutput_callback,
+                dtype="float32",
+                device=self.settings.serverOutputDeviceId,
+                blocksize=block_frame,
+                samplerate=self.settings.serverOutputAudioSampleRate,
+                channels=outputMaxChannel,
+                extra_settings=outputExtraSetting
+            ):
+                while True:
+                    changed = self.checkSettingChanged()
+                    if changed:
+                        break
+                    time.sleep(2)
+                    print(f"[Voice Changer] server audio performance {self.performance}")
+                    print(f"                status: started:{self.settings.serverAudioStated}, model_sr:{self.currentModelSamplingRate}, chunk:{self.currentInputChunkNum}")
+                    print(f"                input  : id:{self.settings.serverInputDeviceId}, sr:{self.settings.serverInputAudioSampleRate}, ch:{inputMaxChannel}")
+                    print(f"                output : id:{self.settings.serverOutputDeviceId}, sr:{self.settings.serverOutputAudioSampleRate}, ch:{outputMaxChannel}")
+                    print(f"                monitor: id:{self.settings.serverMonitorDeviceId}, sr:{self.settings.serverMonitorAudioSampleRate}, ch:{monitorMaxChannel}")
+
+    def runWithMonitorAllSeparate(self, block_frame: int, inputMaxChannel: int, outputMaxChannel: int, monitorMaxChannel: int, inputExtraSetting, outputExtraSetting, monitorExtraSetting):
+        with sd.InputStream(
+            callback=self.audioInput_callback_outQueue_monQueue,
+            dtype="float32",
+            device=self.settings.serverInputDeviceId,
+            blocksize=block_frame,
+            samplerate=self.settings.serverInputAudioSampleRate,
+            channels=inputMaxChannel,
+            extra_settings=inputExtraSetting
+        ):
+            with sd.OutputStream(
+                callback=self.audioOutput_callback,
+                dtype="float32",
+                device=self.settings.serverOutputDeviceId,
+                blocksize=block_frame,
+                samplerate=self.settings.serverOutputAudioSampleRate,
+                channels=outputMaxChannel,
+                extra_settings=outputExtraSetting
+            ):
+                with sd.OutputStream(
+                    callback=self.audioMonitor_callback,
+                    dtype="float32",
+                    device=self.settings.serverMonitorDeviceId,
+                    blocksize=block_frame,
+                    samplerate=self.settings.serverMonitorAudioSampleRate,
+                    channels=monitorMaxChannel,
+                    extra_settings=monitorExtraSetting
+                ):
+                    while True:
+                        changed = self.checkSettingChanged()
+                        if changed:
+                            break
+                        time.sleep(2)
+                        print(f"[Voice Changer] server audio performance {self.performance}")
+                        print(f"                status: started:{self.settings.serverAudioStated}, model_sr:{self.currentModelSamplingRate}, chunk:{self.currentInputChunkNum}")
+                        print(f"                input  : id:{self.settings.serverInputDeviceId}, sr:{self.settings.serverInputAudioSampleRate}, ch:{inputMaxChannel}")
+                        print(f"                output : id:{self.settings.serverOutputDeviceId}, sr:{self.settings.serverOutputAudioSampleRate}, ch:{outputMaxChannel}")
+                        print(f"                monitor: id:{self.settings.serverMonitorDeviceId}, sr:{self.settings.serverMonitorAudioSampleRate}, ch:{monitorMaxChannel}")
+
+    ###########################################
+    # Start Section
+    ###########################################
     def start(self):
-        currentModelSamplingRate = -1
+        self.currentModelSamplingRate = -1
         while True:
             if self.settings.serverAudioStated == 0 or self.settings.serverInputDeviceId == -1:
                 time.sleep(2)
@@ -183,9 +344,9 @@ class ServerDevice:
                 sd._initialize()
 
                 # Curret Device ID
-                currentServerInputDeviceId = self.settings.serverInputDeviceId
+                self.currentServerInputDeviceId = self.settings.serverInputDeviceId
-                currentServerOutputDeviceId = self.settings.serverOutputDeviceId
+                self.currentServerOutputDeviceId = self.settings.serverOutputDeviceId
-                currentServerMonitorDeviceId = self.settings.serverMonitorDeviceId
+                self.currentServerMonitorDeviceId = self.settings.serverMonitorDeviceId
 
                 # Device 特定
                 serverInputAudioDevice = self.getServerInputAudioDevice(self.settings.serverInputDeviceId)
@@ -220,17 +381,17 @@ class ServerDevice:
 
                 # サンプリングレート
                 # 同一サンプリングレートに統一（変換時にサンプルが不足する場合があるため。パディング方法が明らかになれば、それぞれ設定できるかも）
-                currentAudioSampleRate = self.settings.serverAudioSampleRate
+                self.currentAudioSampleRate = self.settings.serverAudioSampleRate
                 try:
-                    currentModelSamplingRate = self.serverDeviceCallbacks.get_processing_sampling_rate()
+                    self.currentModelSamplingRate = self.serverDeviceCallbacks.get_processing_sampling_rate()
                 except Exception as e:
                     print("[Voice Changer] ex: get_processing_sampling_rate", e)
                     time.sleep(2)
                     continue
 
-                self.settings.serverInputAudioSampleRate = currentAudioSampleRate
+                self.settings.serverInputAudioSampleRate = self.currentAudioSampleRate
-                self.settings.serverOutputAudioSampleRate = currentAudioSampleRate
+                self.settings.serverOutputAudioSampleRate = self.currentAudioSampleRate
-                self.settings.serverMonitorAudioSampleRate = currentAudioSampleRate
+                self.settings.serverMonitorAudioSampleRate = self.currentAudioSampleRate
 
                 # Sample Rate Check
                 inputAudioSampleRateAvailable = checkSamplingRate(self.settings.serverInputDeviceId, self.settings.serverInputAudioSampleRate, "input")
@@ -238,7 +399,7 @@ class ServerDevice:
                 monitorAudioSampleRateAvailable = checkSamplingRate(self.settings.serverMonitorDeviceId, self.settings.serverMonitorAudioSampleRate, "output") if serverMonitorAudioDevice else True
 
                 print("Sample Rate:")
-                print(f"  [Model]: {currentModelSamplingRate}")
+                print(f"  [Model]: {self.currentModelSamplingRate}")
                 print(f"  [Input]: {self.settings.serverInputAudioSampleRate} -> {inputAudioSampleRateAvailable}")
                 print(f"  [Output]: {self.settings.serverOutputAudioSampleRate} -> {outputAudioSampleRateAvailable}")
                 if serverMonitorAudioDevice is not None:
@@ -274,153 +435,51 @@ class ServerDevice:
                 self.serverDeviceCallbacks.setOutputSamplingRate(self.settings.serverOutputAudioSampleRate)
 
                 # Blockサイズを計算
-                currentInputChunkNum = self.settings.serverReadChunkSize
+                self.currentInputChunkNum = self.settings.serverReadChunkSize
                 # block_frame = currentInputChunkNum * 128
-                block_frame = int(currentInputChunkNum * 128 * (self.settings.serverInputAudioSampleRate / 48000))
+                block_frame = int(self.currentInputChunkNum * 128 * (self.settings.serverInputAudioSampleRate / 48000))
 
                 sd.default.blocksize = block_frame
 
                 # main loop
                 try:
-                    with sd.InputStream(
+                    # See https://github.com/w-okada/voice-changer/issues/620
-                        callback=self.audioInput_callback,
+                    def judgeServerDeviceMode() -> LocalServerDeviceMode:
-                        dtype="float32",
+                        if self.settings.serverMonitorDeviceId == -1:
-                        device=self.settings.serverInputDeviceId,
+                            return "NoMonitorSeparate"
-                        blocksize=block_frame,
+                        else:
-                        samplerate=self.settings.serverInputAudioSampleRate,
+                            if serverInputAudioDevice.hostAPI == serverOutputAudioDevice.hostAPI and serverInputAudioDevice.hostAPI == serverMonitorAudioDevice.hostAPI:  # すべて同じ
-                        channels=serverInputAudioDevice.maxInputChannels,
+                                return "WithMonitorStandard"
-                        extra_settings=inputExtraSetting
+                            elif serverInputAudioDevice.hostAPI != serverOutputAudioDevice.hostAPI and serverInputAudioDevice.hostAPI != serverMonitorAudioDevice.hostAPI and serverOutputAudioDevice.hostAPI != serverMonitorAudioDevice.hostAPI:  # すべて違う
-                    ):
+                                return "WithMonitorAllSeparate"
-                        with sd.OutputStream(
+                            elif serverInputAudioDevice.hostAPI == serverOutputAudioDevice.hostAPI:  # in/outだけが同じ
-                            callback=self.audioOutput_callback,
+                                return "WithMonitorAllSeparate"
-                            dtype="float32",
+                            elif serverInputAudioDevice.hostAPI == serverMonitorAudioDevice.hostAPI:  # in/monだけが同じ
-                            device=self.settings.serverOutputDeviceId,
+                                return "WithMonitorStandard"
-                            blocksize=block_frame,
+                            elif serverOutputAudioDevice.hostAPI == serverMonitorAudioDevice.hostAPI:  # out/monだけが同じ
-                            samplerate=self.settings.serverOutputAudioSampleRate,
+                                return "WithMonitorAllSeparate"
-                            channels=serverOutputAudioDevice.maxOutputChannels,
-                            extra_settings=outputExtraSetting
-                        ):
-                            if self.settings.serverMonitorDeviceId != -1:
-                                with sd.OutputStream(
-                                    callback=self.audioMonitor_callback,
-                                    dtype="float32",
-                                    device=self.settings.serverMonitorDeviceId,
-                                    blocksize=block_frame,
-                                    samplerate=self.settings.serverMonitorAudioSampleRate,
-                                    channels=serverMonitorAudioDevice.maxOutputChannels,
-                                    extra_settings=monitorExtraSetting
-                                ):
-                                    while (
-                                        self.settings.serverAudioStated == 1 and
-                                        currentServerInputDeviceId == self.settings.serverInputDeviceId and
-                                        currentServerOutputDeviceId == self.settings.serverOutputDeviceId and
-                                        currentServerMonitorDeviceId == self.settings.serverMonitorDeviceId and
-                                        currentModelSamplingRate == self.serverDeviceCallbacks.get_processing_sampling_rate() and
-                                        currentInputChunkNum == self.settings.serverReadChunkSize and
-                                        currentAudioSampleRate == self.settings.serverAudioSampleRate
-                                    ):
-                                        time.sleep(2)
-                                        print(f"[Voice Changer] server audio performance {self.performance}")
-                                        print(f"                status: started:{self.settings.serverAudioStated}, model_sr:{currentModelSamplingRate}, chunk:{currentInputChunkNum}")
-                                        print(f"                input  : id:{self.settings.serverInputDeviceId}, sr:{self.settings.serverInputAudioSampleRate}, ch:{serverInputAudioDevice.maxInputChannels}")
-                                        print(f"                output : id:{self.settings.serverOutputDeviceId}, sr:{self.settings.serverOutputAudioSampleRate}, ch:{serverOutputAudioDevice.maxOutputChannels}")
-                                        print(f"                monitor: id:{self.settings.serverMonitorDeviceId}, sr:{self.settings.serverMonitorAudioSampleRate}, ch:{serverMonitorAudioDevice.maxOutputChannels}")
                             else:
-                                while (
+                                raise RuntimeError(f"Cannot JudgeServerMode, in:{serverInputAudioDevice.hostAPI}, mon:{serverMonitorAudioDevice.hostAPI}, out:{serverOutputAudioDevice.hostAPI}")
-                                    self.settings.serverAudioStated == 1 and
+
-                                    currentServerInputDeviceId == self.settings.serverInputDeviceId and
+                    serverDeviceMode = judgeServerDeviceMode()
-                                    currentServerOutputDeviceId == self.settings.serverOutputDeviceId and
+                    if serverDeviceMode == "NoMonitorSeparate":
-                                    currentServerMonitorDeviceId == self.settings.serverMonitorDeviceId and
+                        self.runNoMonitorSeparate(block_frame, serverInputAudioDevice.maxInputChannels, serverOutputAudioDevice.maxOutputChannels, inputExtraSetting, outputExtraSetting)
-                                    currentModelSamplingRate == self.serverDeviceCallbacks.get_processing_sampling_rate() and
+                    elif serverDeviceMode == "WithMonitorStandard":
-                                    currentInputChunkNum == self.settings.serverReadChunkSize and
+                        self.runWithMonitorStandard(block_frame, serverInputAudioDevice.maxInputChannels, serverOutputAudioDevice.maxOutputChannels, serverMonitorAudioDevice.maxOutputChannels, inputExtraSetting, outputExtraSetting, monitorExtraSetting)
-                                    currentAudioSampleRate == self.settings.serverAudioSampleRate
+                    elif serverDeviceMode == "WithMonitorAllSeparate":
-                                ):
+                        self.runWithMonitorAllSeparate(block_frame, serverInputAudioDevice.maxInputChannels, serverOutputAudioDevice.maxOutputChannels, serverMonitorAudioDevice.maxOutputChannels, inputExtraSetting, outputExtraSetting, monitorExtraSetting)
-                                    time.sleep(2)
+                    else:
-                                    print(f"[Voice Changer] server audio performance {self.performance}")
+                        raise RuntimeError(f"Unknown ServerDeviceMode: {serverDeviceMode}")
-                                    print(f"                status: started:{self.settings.serverAudioStated}, model_sr:{currentModelSamplingRate}, chunk:{currentInputChunkNum}]")
+
-                                    print(f"                input  : id:{self.settings.serverInputDeviceId}, sr:{self.settings.serverInputAudioSampleRate}, ch:{serverInputAudioDevice.maxInputChannels}")
-                                    print(f"                output : id:{self.settings.serverOutputDeviceId}, sr:{self.settings.serverOutputAudioSampleRate}, ch:{serverOutputAudioDevice.maxOutputChannels}")
                 except Exception as e:
                     print("[Voice Changer] processing, ex:", e)
+                    import traceback
+                    traceback.print_exc()
                     time.sleep(2)
 
-    def start2(self):
+    ###########################################
-        # currentInputDeviceId = -1
+    # Info Section
-        # currentOutputDeviceId = -1
+    ###########################################
-        # currentInputChunkNum = -1
-        currentModelSamplingRate = -1
-        while True:
-            if self.settings.serverAudioStated == 0 or self.settings.serverInputDeviceId == -1:
-                time.sleep(2)
-            else:
-                sd._terminate()
-                sd._initialize()
-
-                sd.default.device[0] = self.settings.serverInputDeviceId
-                sd.default.device[1] = self.settings.serverOutputDeviceId
-
-                serverInputAudioDevice = self.getServerInputAudioDevice(sd.default.device[0])
-                serverOutputAudioDevice = self.getServerOutputAudioDevice(sd.default.device[1])
-                print("Devices:", serverInputAudioDevice, serverOutputAudioDevice)
-                if serverInputAudioDevice is None or serverOutputAudioDevice is None:
-                    time.sleep(2)
-                    print("serverInputAudioDevice or serverOutputAudioDevice is None")
-                    continue
-
-                sd.default.channels[0] = serverInputAudioDevice.maxInputChannels
-                sd.default.channels[1] = serverOutputAudioDevice.maxOutputChannels
-
-                currentInputChunkNum = self.settings.serverReadChunkSize
-                block_frame = currentInputChunkNum * 128
-
-                # sample rate precheck(alsa cannot use 40000?)
-                try:
-                    currentModelSamplingRate = self.serverDeviceCallbacks.get_processing_sampling_rate()
-                except Exception as e:
-                    print("[Voice Changer] ex: get_processing_sampling_rate", e)
-                    continue
-                try:
-                    with sd.Stream(
-                        callback=self.audio_callback,
-                        blocksize=block_frame,
-                        # samplerate=currentModelSamplingRate,
-                        dtype="float32",
-                        # dtype="int16",
-                        # channels=[currentInputChannelNum, currentOutputChannelNum],
-                    ):
-                        pass
-                    self.settings.serverInputAudioSampleRate = currentModelSamplingRate
-                    self.serverDeviceCallbacks.setInputSamplingRate(currentModelSamplingRate)
-                    self.serverDeviceCallbacks.setOutputSamplingRate(currentModelSamplingRate)
-                    print(f"[Voice Changer] sample rate {self.settings.serverInputAudioSampleRate}")
-                except Exception as e:
-                    print("[Voice Changer] ex: fallback to device default samplerate", e)
-                    print("[Voice Changer] device default samplerate", serverInputAudioDevice.default_samplerate)
-                    self.settings.serverInputAudioSampleRate = round(serverInputAudioDevice.default_samplerate)
-                    self.serverDeviceCallbacks.setInputSamplingRate(round(serverInputAudioDevice.default_samplerate))
-                    self.serverDeviceCallbacks.setOutputSamplingRate(round(serverInputAudioDevice.default_samplerate))
-
-                sd.default.samplerate = self.settings.serverInputAudioSampleRate
-                sd.default.blocksize = block_frame
-                # main loop
-                try:
-                    with sd.Stream(
-                        callback=self.audio_callback,
-                        # blocksize=block_frame,
-                        # samplerate=vc.settings.serverInputAudioSampleRate,
-                        dtype="float32",
-                        # dtype="int16",
-                        # channels=[currentInputChannelNum, currentOutputChannelNum],
-                    ):
-                        while self.settings.serverAudioStated == 1 and sd.default.device[0] == self.settings.serverInputDeviceId and sd.default.device[1] == self.settings.serverOutputDeviceId and currentModelSamplingRate == self.serverDeviceCallbacks.get_processing_sampling_rate() and currentInputChunkNum == self.settings.serverReadChunkSize:
-                            time.sleep(2)
-                            print("[Voice Changer] server audio", self.performance)
-                            print(f"[Voice Changer] started:{self.settings.serverAudioStated}, input:{sd.default.device[0]}, output:{sd.default.device[1]}, mic_sr:{self.settings.serverInputAudioSampleRate}, model_sr:{currentModelSamplingRate}, chunk:{currentInputChunkNum}, ch:[{sd.default.channels}]")
-
-                except Exception as e:
-                    print("[Voice Changer] ex:", e)
-                    time.sleep(2)
-
     def get_info(self):
         data = asdict(self.settings)
         try:

server/voice_changer/MMVCv13/MMVCv13.py

@@ -1,6 +1,7 @@
 import sys
 import os
 from data.ModelSlot import MMVCv13ModelSlot
+from voice_changer.VoiceChangerParamsManager import VoiceChangerParamsManager
 
 from voice_changer.utils.VoiceChangerModel import AudioInOut
 
@@ -63,19 +64,22 @@ class MMVCv13:
 
     def initialize(self):
         print("[Voice Changer] [MMVCv13] Initializing... ")
+        vcparams = VoiceChangerParamsManager.get_instance().params
+        configPath = os.path.join(vcparams.model_dir, str(self.slotInfo.slotIndex), self.slotInfo.configFile)
+        modelPath = os.path.join(vcparams.model_dir, str(self.slotInfo.slotIndex), self.slotInfo.modelFile)
 
-        self.hps = get_hparams_from_file(self.slotInfo.configFile)
+        self.hps = get_hparams_from_file(configPath)
         if self.slotInfo.isONNX:
             providers, options = self.getOnnxExecutionProvider()
             self.onnx_session = onnxruntime.InferenceSession(
-                self.slotInfo.modelFile,
+                modelPath,
                 providers=providers,
                 provider_options=options,
             )
         else:
             self.net_g = SynthesizerTrn(len(symbols), self.hps.data.filter_length // 2 + 1, self.hps.train.segment_size // self.hps.data.hop_length, n_speakers=self.hps.data.n_speakers, **self.hps.model)
             self.net_g.eval()
-            load_checkpoint(self.slotInfo.modelFile, self.net_g, None)
+            load_checkpoint(modelPath, self.net_g, None)
 
         # その他の設定
         self.settings.srcId = self.slotInfo.srcId
@@ -105,8 +109,10 @@ class MMVCv13:
 
             if key == "gpu" and self.slotInfo.isONNX:
                 providers, options = self.getOnnxExecutionProvider()
+                vcparams = VoiceChangerParamsManager.get_instance().params
+                modelPath = os.path.join(vcparams.model_dir, str(self.slotInfo.slotIndex), self.slotInfo.modelFile)
                 self.onnx_session = onnxruntime.InferenceSession(
-                    self.slotInfo.modelFile,
+                    modelPath,
                     providers=providers,
                     provider_options=options,
                 )
@@ -249,3 +255,15 @@ class MMVCv13:
                     sys.modules.pop(key)
             except:  # NOQA
                 pass
+
+    def get_model_current(self):
+        return [
+            {
+                "key": "srcId",
+                "val": self.settings.srcId,
+            },
+            {
+                "key": "dstId",
+                "val": self.settings.dstId,
+            }
+        ]

server/voice_changer/MMVCv15/MMVCv15.py

@@ -1,6 +1,7 @@
 import sys
 import os
 from data.ModelSlot import MMVCv15ModelSlot
+from voice_changer.VoiceChangerParamsManager import VoiceChangerParamsManager
 from voice_changer.utils.VoiceChangerModel import AudioInOut
 
 if sys.platform.startswith("darwin"):
@@ -70,7 +71,11 @@ class MMVCv15:
 
     def initialize(self):
         print("[Voice Changer] [MMVCv15] Initializing... ")
-        self.hps = get_hparams_from_file(self.slotInfo.configFile)
+        vcparams = VoiceChangerParamsManager.get_instance().params
+        configPath = os.path.join(vcparams.model_dir, str(self.slotInfo.slotIndex), self.slotInfo.configFile)
+        modelPath = os.path.join(vcparams.model_dir, str(self.slotInfo.slotIndex), self.slotInfo.modelFile)
+
+        self.hps = get_hparams_from_file(configPath)
 
         self.net_g = SynthesizerTrn(
             spec_channels=self.hps.data.filter_length // 2 + 1,
@@ -96,7 +101,7 @@ class MMVCv15:
             self.onxx_input_length = 8192
             providers, options = self.getOnnxExecutionProvider()
             self.onnx_session = onnxruntime.InferenceSession(
-                self.slotInfo.modelFile,
+                modelPath,
                 providers=providers,
                 provider_options=options,
             )
@@ -108,7 +113,7 @@ class MMVCv15:
                     self.settings.maxInputLength = self.onxx_input_length - (0.012 * self.hps.data.sampling_rate) - 1024 # onnxの場合は入力長固(crossfadeの1024は仮) # NOQA
         else:
             self.net_g.eval()
-            load_checkpoint(self.slotInfo.modelFile, self.net_g, None)
+            load_checkpoint(modelPath, self.net_g, None)
 
         # その他の設定
         self.settings.srcId = self.slotInfo.srcId
@@ -139,8 +144,10 @@ class MMVCv15:
             setattr(self.settings, key, val)
             if key == "gpu" and self.slotInfo.isONNX:
                 providers, options = self.getOnnxExecutionProvider()
+                vcparams = VoiceChangerParamsManager.get_instance().params
+                modelPath = os.path.join(vcparams.model_dir, str(self.slotInfo.slotIndex), self.slotInfo.modelFile)
                 self.onnx_session = onnxruntime.InferenceSession(
-                    self.slotInfo.modelFile,
+                    modelPath,
                     providers=providers,
                     provider_options=options,
                 )
@@ -208,7 +215,8 @@ class MMVCv15:
         solaSearchFrame: int = 0,
     ):
         # maxInputLength を更新(ここでやると非効率だが、とりあえず。)
-        self.settings.maxInputLength = self.onxx_input_length - crossfadeSize - solaSearchFrame # onnxの場合は入力長固(crossfadeの1024は仮) # NOQA
+        if self.slotInfo.isONNX:
+            self.settings.maxInputLength = self.onxx_input_length - crossfadeSize - solaSearchFrame # onnxの場合は入力長固(crossfadeの1024は仮) # NOQA get_infoで返る値。この関数内の処理では使わない。
 
         newData = newData.astype(np.float32) / self.hps.data.max_wav_value
 
@@ -310,3 +318,19 @@ class MMVCv15:
                     sys.modules.pop(key)
             except: # NOQA
                 pass
+
+    def get_model_current(self):
+        return [
+            {
+                "key": "srcId",
+                "val": self.settings.srcId,
+            },
+            {
+                "key": "dstId",
+                "val": self.settings.dstId,
+            },
+            {
+                "key": "f0Factor",
+                "val": self.settings.f0Factor,
+            }
+        ]

server/voice_changer/MMVCv15/MMVCv15ModelSlotGenerator.py

@@ -1,6 +1,7 @@
 import os
 
 from data.ModelSlot import MMVCv15ModelSlot
+from voice_changer.VoiceChangerParamsManager import VoiceChangerParamsManager
 from voice_changer.utils.LoadModelParams import LoadModelParams
 from voice_changer.utils.ModelSlotGenerator import ModelSlotGenerator
 
@@ -15,7 +16,9 @@ class MMVCv15ModelSlotGenerator(ModelSlotGenerator):
             elif file.kind == "mmvcv15Config":
                 slotInfo.configFile = file.name
             elif file.kind == "mmvcv15Correspondence":
-                with open(file.name, "r") as f:
+                vcparams = VoiceChangerParamsManager.get_instance().params
+                filePath = os.path.join(vcparams.model_dir, str(props.slot), file.name)
+                with open(filePath, "r") as f:
                     slotInfo.speakers = {}
                     while True:
                         line = f.readline()

server/voice_changer/RVC/RVCModelMerger.py

@@ -4,17 +4,17 @@ import torch
 from const import UPLOAD_DIR
 from voice_changer.RVC.modelMerger.MergeModel import merge_model
 from voice_changer.utils.ModelMerger import ModelMerger, ModelMergerRequest
+from voice_changer.utils.VoiceChangerParams import VoiceChangerParams
 
 
 class RVCModelMerger(ModelMerger):
     @classmethod
-    def merge_models(cls, request: ModelMergerRequest, storeSlot: int):
+    def merge_models(cls, params: VoiceChangerParams, request: ModelMergerRequest, storeSlot: int):
-        print("[Voice Changer] MergeRequest:", request)
+        merged = merge_model(params, request)
-        merged = merge_model(request)
 
         # いったんは、アップロードフォルダに格納する。（歴史的経緯）
         # 後続のloadmodelを呼び出すことで永続化モデルフォルダに移動させられる。
-        storeDir = os.path.join(UPLOAD_DIR, f"{storeSlot}")
+        storeDir = os.path.join(UPLOAD_DIR)
         print("[Voice Changer] store merged model to:", storeDir)
         os.makedirs(storeDir, exist_ok=True)
         storeFile = os.path.join(storeDir, "merged.pth")

server/voice_changer/RVC/RVCModelSlotGenerator.py

@@ -5,7 +5,8 @@ import torch
 import onnxruntime
 import json
 
-from data.ModelSlot import ModelSlot, RVCModelSlot
+from data.ModelSlot import RVCModelSlot
+from voice_changer.VoiceChangerParamsManager import VoiceChangerParamsManager
 from voice_changer.utils.LoadModelParams import LoadModelParams
 from voice_changer.utils.ModelSlotGenerator import ModelSlotGenerator
 
@@ -13,6 +14,7 @@ from voice_changer.utils.ModelSlotGenerator import ModelSlotGenerator
 class RVCModelSlotGenerator(ModelSlotGenerator):
     @classmethod
     def loadModel(cls, props: LoadModelParams):
+        vcparams = VoiceChangerParamsManager.get_instance().params
         slotInfo: RVCModelSlot = RVCModelSlot()
         for file in props.files:
             if file.kind == "rvcModel":
@@ -24,17 +26,20 @@ class RVCModelSlotGenerator(ModelSlotGenerator):
         slotInfo.defaultProtect = 0.5
         slotInfo.isONNX = slotInfo.modelFile.endswith(".onnx")
         slotInfo.name = os.path.splitext(os.path.basename(slotInfo.modelFile))[0]
+        print("RVC:: slotInfo.modelFile", slotInfo.modelFile)
+
         # slotInfo.iconFile = "/assets/icons/noimage.png"
 
+        modelPath = os.path.join(vcparams.model_dir, str(props.slot), os.path.basename(slotInfo.modelFile))
         if slotInfo.isONNX:
-            slotInfo = cls._setInfoByONNX(slotInfo)
+            slotInfo = cls._setInfoByONNX(modelPath, slotInfo)
         else:
-            slotInfo = cls._setInfoByPytorch(slotInfo)
+            slotInfo = cls._setInfoByPytorch(modelPath, slotInfo)
         return slotInfo
 
     @classmethod
-    def _setInfoByPytorch(cls, slot: ModelSlot):
+    def _setInfoByPytorch(cls, modelPath: str, slot: RVCModelSlot):
-        cpt = torch.load(slot.modelFile, map_location="cpu")
+        cpt = torch.load(modelPath, map_location="cpu")
         config_len = len(cpt["config"])
         version = cpt.get("version", "v1")
 
@@ -113,8 +118,8 @@ class RVCModelSlotGenerator(ModelSlotGenerator):
         return slot
 
     @classmethod
-    def _setInfoByONNX(cls, slot: ModelSlot):
+    def _setInfoByONNX(cls, modelPath: str, slot: RVCModelSlot):
-        tmp_onnx_session = onnxruntime.InferenceSession(slot.modelFile, providers=["CPUExecutionProvider"])
+        tmp_onnx_session = onnxruntime.InferenceSession(modelPath, providers=["CPUExecutionProvider"])
         modelmeta = tmp_onnx_session.get_modelmeta()
         try:
             slot = RVCModelSlot(**asdict(slot))

server/voice_changer/RVC/RVCr2.py

@@ -0,0 +1,289 @@
+'''
+VoiceChangerV2向け
+'''
+from dataclasses import asdict
+import numpy as np
+import torch
+from data.ModelSlot import RVCModelSlot
+from mods.log_control import VoiceChangaerLogger
+
+from voice_changer.RVC.RVCSettings import RVCSettings
+from voice_changer.RVC.embedder.EmbedderManager import EmbedderManager
+from voice_changer.utils.VoiceChangerModel import AudioInOut, PitchfInOut, FeatureInOut, VoiceChangerModel
+from voice_changer.utils.VoiceChangerParams import VoiceChangerParams
+from voice_changer.RVC.onnxExporter.export2onnx import export2onnx
+from voice_changer.RVC.pitchExtractor.PitchExtractorManager import PitchExtractorManager
+from voice_changer.RVC.pipeline.PipelineGenerator import createPipeline
+from voice_changer.RVC.deviceManager.DeviceManager import DeviceManager
+from voice_changer.RVC.pipeline.Pipeline import Pipeline
+
+from Exceptions import DeviceCannotSupportHalfPrecisionException, PipelineCreateException, PipelineNotInitializedException
+import resampy
+from typing import cast
+
+logger = VoiceChangaerLogger.get_instance().getLogger()
+
+
+class RVCr2(VoiceChangerModel):
+    def __init__(self, params: VoiceChangerParams, slotInfo: RVCModelSlot):
+        logger.info("[Voice Changer] [RVCr2] Creating instance ")
+        self.deviceManager = DeviceManager.get_instance()
+        EmbedderManager.initialize(params)
+        PitchExtractorManager.initialize(params)
+        self.settings = RVCSettings()
+        self.params = params
+        # self.pitchExtractor = PitchExtractorManager.getPitchExtractor(self.settings.f0Detector, self.settings.gpu)
+
+        self.pipeline: Pipeline | None = None
+
+        self.audio_buffer: AudioInOut | None = None
+        self.pitchf_buffer: PitchfInOut | None = None
+        self.feature_buffer: FeatureInOut | None = None
+        self.prevVol = 0.0
+        self.slotInfo = slotInfo
+        # self.initialize()
+
+    def initialize(self):
+        logger.info("[Voice Changer][RVCr2] Initializing... ")
+
+        # pipelineの生成
+        try:
+            self.pipeline = createPipeline(self.params, self.slotInfo, self.settings.gpu, self.settings.f0Detector)
+        except PipelineCreateException as e:  # NOQA
+            logger.error("[Voice Changer] pipeline create failed. check your model is valid.")
+            return
+
+        # その他の設定
+        self.settings.tran = self.slotInfo.defaultTune
+        self.settings.indexRatio = self.slotInfo.defaultIndexRatio
+        self.settings.protect = self.slotInfo.defaultProtect
+        logger.info("[Voice Changer] [RVC] Initializing... done")
+
+    def setSamplingRate(self, inputSampleRate, outputSampleRate):
+        self.inputSampleRate = inputSampleRate
+        self.outputSampleRate = outputSampleRate
+        self.initialize()
+
+    def update_settings(self, key: str, val: int | float | str):
+        logger.info(f"[Voice Changer][RVC]: update_settings {key}:{val}")
+        if key in self.settings.intData:
+            setattr(self.settings, key, int(val))
+            if key == "gpu":
+                self.deviceManager.setForceTensor(False)
+                self.initialize()
+        elif key in self.settings.floatData:
+            setattr(self.settings, key, float(val))
+        elif key in self.settings.strData:
+            setattr(self.settings, key, str(val))
+            if key == "f0Detector" and self.pipeline is not None:
+                pitchExtractor = PitchExtractorManager.getPitchExtractor(self.settings.f0Detector, self.settings.gpu)
+                self.pipeline.setPitchExtractor(pitchExtractor)
+        else:
+            return False
+        return True
+
+    def get_info(self):
+        data = asdict(self.settings)
+        if self.pipeline is not None:
+            pipelineInfo = self.pipeline.getPipelineInfo()
+            data["pipelineInfo"] = pipelineInfo
+        else:
+            data["pipelineInfo"] = "None"
+        return data
+
+    def get_processing_sampling_rate(self):
+        return self.slotInfo.samplingRate
+
+    def generate_input(
+        self,
+        newData: AudioInOut,
+        crossfadeSize: int,
+        solaSearchFrame: int,
+        extra_frame: int
+    ):
+        # 16k で入ってくる。
+        inputSize = newData.shape[0]
+        newData = newData.astype(np.float32) / 32768.0
+        newFeatureLength = inputSize // 160  # hopsize:=160
+
+        if self.audio_buffer is not None:
+            # 過去のデータに連結
+            self.audio_buffer = np.concatenate([self.audio_buffer, newData], 0)
+            if self.slotInfo.f0:
+                self.pitchf_buffer = np.concatenate([self.pitchf_buffer, np.zeros(newFeatureLength)], 0)
+            self.feature_buffer = np.concatenate([self.feature_buffer, np.zeros([newFeatureLength, self.slotInfo.embChannels])], 0)
+        else:
+            self.audio_buffer = newData
+            if self.slotInfo.f0:
+                self.pitchf_buffer = np.zeros(newFeatureLength)
+            self.feature_buffer = np.zeros([newFeatureLength, self.slotInfo.embChannels])
+
+        convertSize = inputSize + crossfadeSize + solaSearchFrame + extra_frame
+
+        if convertSize % 160 != 0:  # モデルの出力のホップサイズで切り捨てが発生するので補う。
+            convertSize = convertSize + (160 - (convertSize % 160))
+        outSize = int(((convertSize - extra_frame) / 16000) * self.slotInfo.samplingRate) 
+
+        # バッファがたまっていない場合はzeroで補う
+        if self.audio_buffer.shape[0] < convertSize:
+            self.audio_buffer = np.concatenate([np.zeros([convertSize]), self.audio_buffer])
+            if self.slotInfo.f0:
+                self.pitchf_buffer = np.concatenate([np.zeros([convertSize // 160]), self.pitchf_buffer])
+            self.feature_buffer = np.concatenate([np.zeros([convertSize // 160, self.slotInfo.embChannels]), self.feature_buffer])
+
+        # 不要部分をトリミング
+        convertOffset = -1 * convertSize
+        featureOffset = convertOffset // 160
+        self.audio_buffer = self.audio_buffer[convertOffset:]  # 変換対象の部分だけ抽出
+        if self.slotInfo.f0:
+            self.pitchf_buffer = self.pitchf_buffer[featureOffset:]
+        self.feature_buffer = self.feature_buffer[featureOffset:]
+
+        # 出力部分だけ切り出して音量を確認。(TODO:段階的消音にする)
+        cropOffset = -1 * (inputSize + crossfadeSize)
+        cropEnd = -1 * (crossfadeSize)
+        crop = self.audio_buffer[cropOffset:cropEnd]
+        vol = np.sqrt(np.square(crop).mean())
+        vol = max(vol, self.prevVol * 0.0)
+        self.prevVol = vol
+
+        return (self.audio_buffer, self.pitchf_buffer, self.feature_buffer, convertSize, vol, outSize)
+
+    def inference(self, receivedData: AudioInOut, crossfade_frame: int, sola_search_frame: int):
+        if self.pipeline is None:
+            logger.info("[Voice Changer] Pipeline is not initialized.")
+            raise PipelineNotInitializedException()
+
+        # 処理は16Kで実施(Pitch, embed, (infer))
+        receivedData = cast(
+            AudioInOut,
+            resampy.resample(
+                receivedData,
+                self.inputSampleRate,
+                16000,
+            ),
+        )
+        crossfade_frame = int((crossfade_frame / self.inputSampleRate) * 16000)
+        sola_search_frame = int((sola_search_frame / self.inputSampleRate) * 16000)
+        extra_frame = int((self.settings.extraConvertSize / self.inputSampleRate) * 16000)
+
+        # 入力データ生成
+        data = self.generate_input(receivedData, crossfade_frame, sola_search_frame, extra_frame)
+
+        audio = data[0]
+        pitchf = data[1]
+        feature = data[2]
+        convertSize = data[3]
+        vol = data[4]
+        outSize = data[5]
+
+        if vol < self.settings.silentThreshold:
+            return np.zeros(convertSize).astype(np.int16) * np.sqrt(vol)
+
+        device = self.pipeline.device
+        
+        audio = torch.from_numpy(audio).to(device=device, dtype=torch.float32)
+        repeat = 1 if self.settings.rvcQuality else 0
+        sid = self.settings.dstId
+        f0_up_key = self.settings.tran
+        index_rate = self.settings.indexRatio
+        protect = self.settings.protect
+
+        if_f0 = 1 if self.slotInfo.f0 else 0
+        embOutputLayer = self.slotInfo.embOutputLayer
+        useFinalProj = self.slotInfo.useFinalProj
+        
+        try:
+            audio_out, self.pitchf_buffer, self.feature_buffer = self.pipeline.exec(
+                sid,
+                audio,
+                pitchf,
+                feature,
+                f0_up_key,
+                index_rate,
+                if_f0,
+                # 0,
+                self.settings.extraConvertSize / self.inputSampleRate if self.settings.silenceFront else 0.,  # extaraDataSizeの秒数。入力のサンプリングレートで算出
+                embOutputLayer,
+                useFinalProj,
+                repeat,
+                protect,
+                outSize
+            )
+            # result = audio_out.detach().cpu().numpy() * np.sqrt(vol)
+            result = audio_out[-outSize:].detach().cpu().numpy() * np.sqrt(vol)
+
+            result = cast(
+                AudioInOut,
+                resampy.resample(
+                    result,
+                    self.slotInfo.samplingRate,
+                    self.outputSampleRate,
+                ),
+            )
+
+            return result
+        except DeviceCannotSupportHalfPrecisionException as e:  # NOQA
+            logger.warn("[Device Manager] Device cannot support half precision. Fallback to float....")
+            self.deviceManager.setForceTensor(True)
+            self.initialize()
+            # raise e
+
+        return
+
+    def __del__(self):
+        del self.pipeline
+
+        # print("---------- REMOVING ---------------")
+
+        # remove_path = os.path.join("RVC")
+        # sys.path = [x for x in sys.path if x.endswith(remove_path) is False]
+
+        # for key in list(sys.modules):
+        #     val = sys.modules.get(key)
+        #     try:
+        #         file_path = val.__file__
+        #         if file_path.find("RVC" + os.path.sep) >= 0:
+        #             # print("remove", key, file_path)
+        #             sys.modules.pop(key)
+        #     except Exception:  # type:ignore
+        #         # print(e)
+        #         pass
+
+    def export2onnx(self):
+        modelSlot = self.slotInfo
+
+        if modelSlot.isONNX:
+            logger.warn("[Voice Changer] export2onnx, No pyTorch filepath.")
+            return {"status": "ng", "path": ""}
+
+        if self.pipeline is not None:
+            del self.pipeline
+            self.pipeline = None
+
+        torch.cuda.empty_cache()
+        self.initialize()
+
+        output_file_simple = export2onnx(self.settings.gpu, modelSlot)
+
+        return {
+            "status": "ok",
+            "path": f"/tmp/{output_file_simple}",
+            "filename": output_file_simple,
+        }
+
+    def get_model_current(self):
+        return [
+            {
+                "key": "defaultTune",
+                "val": self.settings.tran,
+            },
+            {
+                "key": "defaultIndexRatio",
+                "val": self.settings.indexRatio,
+            },
+            {
+                "key": "defaultProtect",
+                "val": self.settings.protect,
+            },
+        ]

server/voice_changer/RVC/embedder/EmbedderManager.py

@@ -46,7 +46,7 @@ class EmbedderManager:
                 file = cls.params.content_vec_500_onnx
                 return OnnxContentvec().loadModel(file, dev)
             except Exception as e:  # noqa
-                print("[Voice Changer] use torch contentvec")
+                print("[Voice Changer] use torch contentvec", e)
                 file = cls.params.hubert_base
                 return FairseqHubert().loadModel(file, dev, isHalf)
         elif embederType == "hubert-base-japanese":

server/voice_changer/RVC/inferencer/InferencerManager.py

@@ -8,7 +8,7 @@ from voice_changer.RVC.inferencer.RVCInferencerv2 import RVCInferencerv2
 from voice_changer.RVC.inferencer.RVCInferencerv2Nono import RVCInferencerv2Nono
 from voice_changer.RVC.inferencer.WebUIInferencer import WebUIInferencer
 from voice_changer.RVC.inferencer.WebUIInferencerNono import WebUIInferencerNono
-from voice_changer.RVC.inferencer.VorasInferencebeta import VoRASInferencer
+import sys
 
 
 class InferencerManager:
@@ -38,7 +38,11 @@ class InferencerManager:
         elif inferencerType == EnumInferenceTypes.pyTorchRVCv2 or inferencerType == EnumInferenceTypes.pyTorchRVCv2.value:
             return RVCInferencerv2().loadModel(file, gpu)
         elif inferencerType == EnumInferenceTypes.pyTorchVoRASbeta or inferencerType == EnumInferenceTypes.pyTorchVoRASbeta.value:
-            return VoRASInferencer().loadModel(file, gpu)
+            if sys.platform.startswith("darwin") is False:
+                from voice_changer.RVC.inferencer.VorasInferencebeta import VoRASInferencer
+                return VoRASInferencer().loadModel(file, gpu)
+            else:
+                raise RuntimeError("[Voice Changer] VoRAS is not supported on macOS")
         elif inferencerType == EnumInferenceTypes.pyTorchRVCv2Nono or inferencerType == EnumInferenceTypes.pyTorchRVCv2Nono.value:
             return RVCInferencerv2Nono().loadModel(file, gpu)
         elif inferencerType == EnumInferenceTypes.pyTorchWebUI or inferencerType == EnumInferenceTypes.pyTorchWebUI.value:

server/voice_changer/RVC/modelMerger/MergeModel.py

@@ -1,12 +1,14 @@
 from typing import Dict, Any
-
+import os
 from collections import OrderedDict
 import torch
+from voice_changer.ModelSlotManager import ModelSlotManager
 
 from voice_changer.utils.ModelMerger import ModelMergerRequest
+from voice_changer.utils.VoiceChangerParams import VoiceChangerParams
 
 
-def merge_model(request: ModelMergerRequest):
+def merge_model(params: VoiceChangerParams, request: ModelMergerRequest):
     def extract(ckpt: Dict[str, Any]):
         a = ckpt["model"]
         opt: Dict[str, Any] = OrderedDict()
@@ -34,11 +36,16 @@ def merge_model(request: ModelMergerRequest):
 
     weights = []
     alphas = []
+    slotManager = ModelSlotManager.get_instance(params.model_dir)
     for f in files:
         strength = f.strength
         if strength == 0:
             continue
-        weight, state_dict = load_weight(f.filename)
+        slotInfo = slotManager.get_slot_info(f.slotIndex)
+
+        filename = os.path.join(params.model_dir, str(f.slotIndex), os.path.basename(slotInfo.modelFile))  # slotInfo.modelFileはv.1.5.3.11以前はmodel_dirから含まれている。
+
+        weight, state_dict = load_weight(filename)
         weights.append(weight)
         alphas.append(f.strength)
 

server/voice_changer/RVC/onnxExporter/export2onnx.py

@@ -4,7 +4,7 @@ import torch
 from onnxsim import simplify
 import onnx
 from const import TMP_DIR, EnumInferenceTypes
-from data.ModelSlot import ModelSlot
+from data.ModelSlot import RVCModelSlot
 from voice_changer.RVC.deviceManager.DeviceManager import DeviceManager
 from voice_changer.RVC.onnxExporter.SynthesizerTrnMs256NSFsid_ONNX import (
     SynthesizerTrnMs256NSFsid_ONNX,
@@ -24,10 +24,12 @@ from voice_changer.RVC.onnxExporter.SynthesizerTrnMsNSFsidNono_webui_ONNX import
 from voice_changer.RVC.onnxExporter.SynthesizerTrnMsNSFsid_webui_ONNX import (
     SynthesizerTrnMsNSFsid_webui_ONNX,
 )
+from voice_changer.VoiceChangerParamsManager import VoiceChangerParamsManager
 
 
-def export2onnx(gpu: int, modelSlot: ModelSlot):
+def export2onnx(gpu: int, modelSlot: RVCModelSlot):
-    modelFile = modelSlot.modelFile
+    vcparams = VoiceChangerParamsManager.get_instance().params
+    modelFile = os.path.join(vcparams.model_dir, str(modelSlot.slotIndex), os.path.basename(modelSlot.modelFile))
 
     output_file = os.path.splitext(os.path.basename(modelFile))[0] + ".onnx"
     output_file_simple = os.path.splitext(os.path.basename(modelFile))[0] + "_simple.onnx"

server/voice_changer/RVC/pipeline/Pipeline.py

@@ -18,6 +18,7 @@ from voice_changer.RVC.inferencer.OnnxRVCInferencer import OnnxRVCInferencer
 from voice_changer.RVC.inferencer.OnnxRVCInferencerNono import OnnxRVCInferencerNono
 
 from voice_changer.RVC.pitchExtractor.PitchExtractor import PitchExtractor
+from voice_changer.utils.Timer import Timer
 
 logger = VoiceChangaerLogger.get_instance().getLogger()
 
@@ -89,174 +90,181 @@ class Pipeline(object):
         protect=0.5,
         out_size=None,
     ):
-        # 16000のサンプリングレートで入ってきている。以降この世界は16000で処理。
+        # print(f"pipeline exec input, audio:{audio.shape}, pitchf:{pitchf.shape}, feature:{feature.shape}")
+        # print(f"pipeline exec input, silence_front:{silence_front}, out_size:{out_size}")
 
-        search_index = self.index is not None and self.big_npy is not None and index_rate != 0
+        with Timer("main-process", False) as t:  # NOQA
-        # self.t_pad = self.sr * repeat  # 1秒
+            # 16000のサンプリングレートで入ってきている。以降この世界は16000で処理。
-        # self.t_pad_tgt = self.targetSR * repeat  # 1秒　出力時のトリミング(モデルのサンプリングで出力される)
+            search_index = self.index is not None and self.big_npy is not None and index_rate != 0
-        audio = audio.unsqueeze(0)
+            # self.t_pad = self.sr * repeat  # 1秒
+            # self.t_pad_tgt = self.targetSR * repeat  # 1秒　出力時のトリミング(モデルのサンプリングで出力される)
+            audio = audio.unsqueeze(0)
 
-        quality_padding_sec = (repeat * (audio.shape[1] - 1)) / self.sr  # padding(reflect)のサイズは元のサイズより小さい必要がある。
+            quality_padding_sec = (repeat * (audio.shape[1] - 1)) / self.sr  # padding(reflect)のサイズは元のサイズより小さい必要がある。
 
-        self.t_pad = round(self.sr * quality_padding_sec)  # 前後に音声を追加
+            self.t_pad = round(self.sr * quality_padding_sec)  # 前後に音声を追加
-        self.t_pad_tgt = round(self.targetSR * quality_padding_sec)  # 前後に音声を追加　出力時のトリミング(モデルのサンプリングで出力される)
+            self.t_pad_tgt = round(self.targetSR * quality_padding_sec)  # 前後に音声を追加　出力時のトリミング(モデルのサンプリングで出力される)
-        audio_pad = F.pad(audio, (self.t_pad, self.t_pad), mode="reflect").squeeze(0)
+            audio_pad = F.pad(audio, (self.t_pad, self.t_pad), mode="reflect").squeeze(0)
-        p_len = audio_pad.shape[0] // self.window
+            p_len = audio_pad.shape[0] // self.window
-        sid = torch.tensor(sid, device=self.device).unsqueeze(0).long()
+            sid = torch.tensor(sid, device=self.device).unsqueeze(0).long()
 
-        # RVC QualityがOnのときにはsilence_frontをオフに。
+            # RVC QualityがOnのときにはsilence_frontをオフに。
-        silence_front = silence_front if repeat == 0 else 0
+            silence_front = silence_front if repeat == 0 else 0
-        pitchf = pitchf if repeat == 0 else np.zeros(p_len)
+            pitchf = pitchf if repeat == 0 else np.zeros(p_len)
-        out_size = out_size if repeat == 0 else None
+            out_size = out_size if repeat == 0 else None
 
-        # ピッチ検出
+            # ピッチ検出
-        try:
-            if if_f0 == 1:
-                pitch, pitchf = self.pitchExtractor.extract(
-                    audio_pad,
-                    pitchf,
-                    f0_up_key,
-                    self.sr,
-                    self.window,
-                    silence_front=silence_front,
-                )
-                # pitch = pitch[:p_len]
-                # pitchf = pitchf[:p_len]
-                pitch = torch.tensor(pitch, device=self.device).unsqueeze(0).long()
-                pitchf = torch.tensor(pitchf, device=self.device, dtype=torch.float).unsqueeze(0)
-            else:
-                pitch = None
-                pitchf = None
-        except IndexError as e:  # NOQA
-            # print(e)
-            # import traceback
-            # traceback.print_exc()
-            raise NotEnoughDataExtimateF0()
-
-        # tensor型調整
-        feats = audio_pad
-        if feats.dim() == 2:  # double channels
-            feats = feats.mean(-1)
-        assert feats.dim() == 1, feats.dim()
-        feats = feats.view(1, -1)
-
-        # embedding
-        with autocast(enabled=self.isHalf):
             try:
-                feats = self.embedder.extractFeatures(feats, embOutputLayer, useFinalProj)
+                if if_f0 == 1:
-                if torch.isnan(feats).all():
+                    pitch, pitchf = self.pitchExtractor.extract(
-                    raise DeviceCannotSupportHalfPrecisionException()
+                        audio_pad,
+                        pitchf,
+                        f0_up_key,
+                        self.sr,
+                        self.window,
+                        silence_front=silence_front,
+                    )
+                    # pitch = pitch[:p_len]
+                    # pitchf = pitchf[:p_len]
+                    pitch = torch.tensor(pitch, device=self.device).unsqueeze(0).long()
+                    pitchf = torch.tensor(pitchf, device=self.device, dtype=torch.float).unsqueeze(0)
+                else:
+                    pitch = None
+                    pitchf = None
+            except IndexError as e:  # NOQA
+                # print(e)
+                # import traceback
+                # traceback.print_exc()
+                raise NotEnoughDataExtimateF0()
+
+            # tensor型調整
+            feats = audio_pad
+            if feats.dim() == 2:  # double channels
+                feats = feats.mean(-1)
+            assert feats.dim() == 1, feats.dim()
+            feats = feats.view(1, -1)
+
+            # embedding
+            with Timer("main-process", False) as te:
+                with autocast(enabled=self.isHalf):
+                    try:
+                        feats = self.embedder.extractFeatures(feats, embOutputLayer, useFinalProj)
+                        if torch.isnan(feats).all():
+                            raise DeviceCannotSupportHalfPrecisionException()
+                    except RuntimeError as e:
+                        if "HALF" in e.__str__().upper():
+                            raise HalfPrecisionChangingException()
+                        elif "same device" in e.__str__():
+                            raise DeviceChangingException()
+                        else:
+                            raise e
+            # print(f"[Embedding] {te.secs}")
+
+            # Index - feature抽出
+            # if self.index is not None and self.feature is not None and index_rate != 0:
+            if search_index:
+                npy = feats[0].cpu().numpy()
+                # apply silent front for indexsearch
+                npyOffset = math.floor(silence_front * 16000) // 360
+                npy = npy[npyOffset:]
+
+                if self.isHalf is True:
+                    npy = npy.astype("float32")
+
+                # TODO: kは調整できるようにする
+                k = 1
+                if k == 1:
+                    _, ix = self.index.search(npy, 1)
+                    npy = self.big_npy[ix.squeeze()]
+                else:
+                    score, ix = self.index.search(npy, k=8)
+                    weight = np.square(1 / score)
+                    weight /= weight.sum(axis=1, keepdims=True)
+                    npy = np.sum(self.big_npy[ix] * np.expand_dims(weight, axis=2), axis=1)
+
+                # recover silient font
+                npy = np.concatenate([np.zeros([npyOffset, npy.shape[1]], dtype=np.float32), feature[:npyOffset:2].astype("float32"), npy])[-feats.shape[1]:]
+                feats = torch.from_numpy(npy).unsqueeze(0).to(self.device) * index_rate + (1 - index_rate) * feats
+            feats = F.interpolate(feats.permute(0, 2, 1), scale_factor=2).permute(0, 2, 1)
+            if protect < 0.5 and search_index:
+                feats0 = feats.clone()
+
+            # ピッチサイズ調整
+            p_len = audio_pad.shape[0] // self.window
+            if feats.shape[1] < p_len:
+                p_len = feats.shape[1]
+                if pitch is not None and pitchf is not None:
+                    pitch = pitch[:, :p_len]
+                    pitchf = pitchf[:, :p_len]
+
+            feats_len = feats.shape[1]
+            if pitch is not None and pitchf is not None:
+                pitch = pitch[:, -feats_len:]
+                pitchf = pitchf[:, -feats_len:]
+            p_len = torch.tensor([feats_len], device=self.device).long()
+
+            # pitchの推定が上手くいかない(pitchf=0)場合、検索前の特徴を混ぜる
+            # pitchffの作り方の疑問はあるが、本家通りなので、このまま使うことにする。
+            # https://github.com/w-okada/voice-changer/pull/276#issuecomment-1571336929
+            if protect < 0.5 and search_index:
+                pitchff = pitchf.clone()
+                pitchff[pitchf > 0] = 1
+                pitchff[pitchf < 1] = protect
+                pitchff = pitchff.unsqueeze(-1)
+                feats = feats * pitchff + feats0 * (1 - pitchff)
+                feats = feats.to(feats0.dtype)
+            p_len = torch.tensor([p_len], device=self.device).long()
+
+            # apply silent front for inference
+            if type(self.inferencer) in [OnnxRVCInferencer, OnnxRVCInferencerNono]:
+                npyOffset = math.floor(silence_front * 16000) // 360
+                feats = feats[:, npyOffset * 2 :, :]  # NOQA
+
+            feats_len = feats.shape[1]
+            if pitch is not None and pitchf is not None:
+                pitch = pitch[:, -feats_len:]
+                pitchf = pitchf[:, -feats_len:]
+            p_len = torch.tensor([feats_len], device=self.device).long()
+
+            # 推論実行
+            try:
+                with torch.no_grad():
+                    with autocast(enabled=self.isHalf):
+                        audio1 = (
+                            torch.clip(
+                                self.inferencer.infer(feats, p_len, pitch, pitchf, sid, out_size)[0][0, 0].to(dtype=torch.float32),
+                                -1.0,
+                                1.0,
+                            )
+                            * 32767.5
+                        ).data.to(dtype=torch.int16)
             except RuntimeError as e:
                 if "HALF" in e.__str__().upper():
+                    print("11", e)
                     raise HalfPrecisionChangingException()
-                elif "same device" in e.__str__():
-                    raise DeviceChangingException()
                 else:
                     raise e
 
-        # Index - feature抽出
+            feats_buffer = feats.squeeze(0).detach().cpu()
-        # if self.index is not None and self.feature is not None and index_rate != 0:
-        if search_index:
-            npy = feats[0].cpu().numpy()
-            # apply silent front for indexsearch
-            npyOffset = math.floor(silence_front * 16000) // 360
-            npy = npy[npyOffset:]
 
-            if self.isHalf is True:
+            if pitchf is not None:
-                npy = npy.astype("float32")
+                pitchf_buffer = pitchf.squeeze(0).detach().cpu()
-
-            # TODO: kは調整できるようにする
-            k = 1
-            if k == 1:
-                _, ix = self.index.search(npy, 1)
-                npy = self.big_npy[ix.squeeze()]
             else:
-                score, ix = self.index.search(npy, k=8)
+                pitchf_buffer = None
-                weight = np.square(1 / score)
-                weight /= weight.sum(axis=1, keepdims=True)
-                npy = np.sum(self.big_npy[ix] * np.expand_dims(weight, axis=2), axis=1)
 
-            # recover silient font
+            del p_len, pitch, pitchf, feats
-            npy = np.concatenate([np.zeros([npyOffset, npy.shape[1]], dtype=np.float32), feature[:npyOffset:2].astype("float32"), npy])[-feats.shape[1]:]
+            # torch.cuda.empty_cache()
-            feats = torch.from_numpy(npy).unsqueeze(0).to(self.device) * index_rate + (1 - index_rate) * feats
-        feats = F.interpolate(feats.permute(0, 2, 1), scale_factor=2).permute(0, 2, 1)
-        if protect < 0.5 and search_index:
-            feats0 = feats.clone()
 
-        # ピッチサイズ調整
+            # inferで出力されるサンプリングレートはモデルのサンプリングレートになる。
-        p_len = audio_pad.shape[0] // self.window
+            # pipelineに（入力されるときはhubertように16k）
-        if feats.shape[1] < p_len:
+            if self.t_pad_tgt != 0:
-            p_len = feats.shape[1]
+                offset = self.t_pad_tgt
-            if pitch is not None and pitchf is not None:
+                end = -1 * self.t_pad_tgt
-                pitch = pitch[:, :p_len]
+                audio1 = audio1[offset:end]
-                pitchf = pitchf[:, :p_len]
 
-        feats_len = feats.shape[1]
+            del sid
-        if pitch is not None and pitchf is not None:
+            # torch.cuda.empty_cache()
-            pitch = pitch[:, -feats_len:]
+        # print("EXEC AVERAGE:", t.avrSecs)
-            pitchf = pitchf[:, -feats_len:]
-        p_len = torch.tensor([feats_len], device=self.device).long()
-
-        # pitchの推定が上手くいかない(pitchf=0)場合、検索前の特徴を混ぜる
-        # pitchffの作り方の疑問はあるが、本家通りなので、このまま使うことにする。
-        # https://github.com/w-okada/voice-changer/pull/276#issuecomment-1571336929
-        if protect < 0.5 and search_index:
-            pitchff = pitchf.clone()
-            pitchff[pitchf > 0] = 1
-            pitchff[pitchf < 1] = protect
-            pitchff = pitchff.unsqueeze(-1)
-            feats = feats * pitchff + feats0 * (1 - pitchff)
-            feats = feats.to(feats0.dtype)
-        p_len = torch.tensor([p_len], device=self.device).long()
-
-        # apply silent front for inference
-        if type(self.inferencer) in [OnnxRVCInferencer, OnnxRVCInferencerNono]:
-            npyOffset = math.floor(silence_front * 16000) // 360
-            feats = feats[:, npyOffset * 2 :, :]  # NOQA
-
-        feats_len = feats.shape[1]
-        if pitch is not None and pitchf is not None:
-            pitch = pitch[:, -feats_len:]
-            pitchf = pitchf[:, -feats_len:]
-        p_len = torch.tensor([feats_len], device=self.device).long()
-
-        # 推論実行
-        try:
-            with torch.no_grad():
-                with autocast(enabled=self.isHalf):
-                    audio1 = (
-                        torch.clip(
-                            self.inferencer.infer(feats, p_len, pitch, pitchf, sid, out_size)[0][0, 0].to(dtype=torch.float32),
-                            -1.0,
-                            1.0,
-                        )
-                        * 32767.5
-                    ).data.to(dtype=torch.int16)
-        except RuntimeError as e:
-            if "HALF" in e.__str__().upper():
-                print("11", e)
-                raise HalfPrecisionChangingException()
-            else:
-                raise e
-
-        feats_buffer = feats.squeeze(0).detach().cpu()
-        if pitchf is not None:
-            pitchf_buffer = pitchf.squeeze(0).detach().cpu()
-        else:
-            pitchf_buffer = None
-
-        del p_len, pitch, pitchf, feats
-        # torch.cuda.empty_cache()
-
-        # inferで出力されるサンプリングレートはモデルのサンプリングレートになる。
-        # pipelineに（入力されるときはhubertように16k）
-        if self.t_pad_tgt != 0:
-            offset = self.t_pad_tgt
-            end = -1 * self.t_pad_tgt
-            audio1 = audio1[offset:end]
-
-        del sid
-        # torch.cuda.empty_cache()
         return audio1, pitchf_buffer, feats_buffer
 
     def __del__(self):

server/voice_changer/RVC/pipeline/PipelineGenerator.py

@@ -9,15 +9,17 @@ from voice_changer.RVC.embedder.EmbedderManager import EmbedderManager
 from voice_changer.RVC.inferencer.InferencerManager import InferencerManager
 from voice_changer.RVC.pipeline.Pipeline import Pipeline
 from voice_changer.RVC.pitchExtractor.PitchExtractorManager import PitchExtractorManager
+from voice_changer.utils.VoiceChangerParams import VoiceChangerParams
 
 
-def createPipeline(modelSlot: RVCModelSlot, gpu: int, f0Detector: str):
+def createPipeline(params: VoiceChangerParams, modelSlot: RVCModelSlot, gpu: int, f0Detector: str):
     dev = DeviceManager.get_instance().getDevice(gpu)
     half = DeviceManager.get_instance().halfPrecisionAvailable(gpu)
 
     # Inferencer 生成
     try:
-        inferencer = InferencerManager.getInferencer(modelSlot.modelType, modelSlot.modelFile, gpu)
+        modelPath = os.path.join(params.model_dir, str(modelSlot.slotIndex), os.path.basename(modelSlot.modelFile))
+        inferencer = InferencerManager.getInferencer(modelSlot.modelType, modelPath, gpu)
     except Exception as e:
         print("[Voice Changer] exception! loading inferencer", e)
         traceback.print_exc()
@@ -40,7 +42,8 @@ def createPipeline(modelSlot: RVCModelSlot, gpu: int, f0Detector: str):
     pitchExtractor = PitchExtractorManager.getPitchExtractor(f0Detector, gpu)
 
     # index, feature
-    index = _loadIndex(modelSlot)
+    indexPath = os.path.join(params.model_dir, str(modelSlot.slotIndex), os.path.basename(modelSlot.indexFile))
+    index = _loadIndex(indexPath)
 
     pipeline = Pipeline(
         embedder,
@@ -55,21 +58,17 @@ def createPipeline(modelSlot: RVCModelSlot, gpu: int, f0Detector: str):
     return pipeline
 
 
-def _loadIndex(modelSlot: RVCModelSlot):
+def _loadIndex(indexPath: str):
     # Indexのロード
     print("[Voice Changer] Loading index...")
-    # ファイル指定がない場合はNone
-    if modelSlot.indexFile is None:
-        print("[Voice Changer] Index is None, not used")
-        return None
-
     # ファイル指定があってもファイルがない場合はNone
-    if os.path.exists(modelSlot.indexFile) is not True:
+    if os.path.exists(indexPath) is not True or os.path.isfile(indexPath) is not True:
+        print("[Voice Changer] Index file is not found")
         return None
 
     try:
-        print("Try loading...", modelSlot.indexFile)
+        print("Try loading...", indexPath)
-        index = faiss.read_index(modelSlot.indexFile)
+        index = faiss.read_index(indexPath)
     except: # NOQA
         print("[Voice Changer] load index failed. Use no index.")
         traceback.print_exc()

server/voice_changer/SoVitsSvc40/SoVitsSvc40.py

@@ -1,6 +1,7 @@
 import sys
 import os
 from data.ModelSlot import SoVitsSvc40ModelSlot
+from voice_changer.VoiceChangerParamsManager import VoiceChangerParamsManager
 
 from voice_changer.utils.VoiceChangerModel import AudioInOut
 from voice_changer.utils.VoiceChangerParams import VoiceChangerParams
@@ -92,13 +93,17 @@ class SoVitsSvc40:
 
     def initialize(self):
         print("[Voice Changer] [so-vits-svc40] Initializing... ")
-        self.hps = get_hparams_from_file(self.slotInfo.configFile)
+        vcparams = VoiceChangerParamsManager.get_instance().params
+        configPath = os.path.join(vcparams.model_dir, str(self.slotInfo.slotIndex), self.slotInfo.configFile)
+        modelPath = os.path.join(vcparams.model_dir, str(self.slotInfo.slotIndex), self.slotInfo.modelFile)
+        self.hps = get_hparams_from_file(configPath)
         self.settings.speakers = self.hps.spk
 
         # cluster
         try:
             if self.slotInfo.clusterFile is not None:
-                self.cluster_model = get_cluster_model(self.slotInfo.clusterFile)
+                clusterPath = os.path.join(vcparams.model_dir, str(self.slotInfo.slotIndex), self.slotInfo.clusterFile)
+                self.cluster_model = get_cluster_model(clusterPath)
             else:
                 self.cluster_model = None
         except Exception as e:
@@ -110,7 +115,7 @@ class SoVitsSvc40:
         if self.slotInfo.isONNX:
             providers, options = self.getOnnxExecutionProvider()
             self.onnx_session = onnxruntime.InferenceSession(
-                self.slotInfo.modelFile,
+                modelPath,
                 providers=providers,
                 provider_options=options,
             )
@@ -122,7 +127,7 @@ class SoVitsSvc40:
             )
             net_g.eval()
             self.net_g = net_g
-            load_checkpoint(self.slotInfo.modelFile, self.net_g, None)
+            load_checkpoint(modelPath, self.net_g, None)
 
     def getOnnxExecutionProvider(self):
         availableProviders = onnxruntime.get_available_providers()
@@ -379,6 +384,10 @@ class SoVitsSvc40:
             except Exception:  # type:ignore
                 pass
 
+    def get_model_current(self):
+        return [
+        ]
+
 
 def resize_f0(x, target_len):
     source = np.array(x)

server/voice_changer/VoiceChangerManager.py

@@ -37,8 +37,14 @@ class GPUInfo:
 @dataclass()
 class VoiceChangerManagerSettings:
     modelSlotIndex: int = -1
+    passThrough: bool = False  # 0: off, 1: on
     # ↓mutableな物だけ列挙
-    intData: list[str] = field(default_factory=lambda: ["modelSlotIndex"])
+    boolData: list[str] = field(default_factory=lambda: [
+        "passThrough"
+    ])
+    intData: list[str] = field(default_factory=lambda: [
+        "modelSlotIndex",
+    ])
 
 
 class VoiceChangerManager(ServerDeviceCallbacks):
@@ -121,7 +127,6 @@ class VoiceChangerManager(ServerDeviceCallbacks):
     def get_instance(cls, params: VoiceChangerParams):
         if cls._instance is None:
             cls._instance = cls(params)
-            # cls._instance.voiceChanger = VoiceChanger(params)
         return cls._instance
 
     def loadModel(self, params: LoadModelParams):
@@ -147,7 +152,7 @@ class VoiceChangerManager(ServerDeviceCallbacks):
                 os.makedirs(dstDir, exist_ok=True)
                 logger.info(f"move to {srcPath} -> {dstPath}")
                 shutil.move(srcPath, dstPath)
-                file.name = dstPath
+                file.name = os.path.basename(dstPath)
 
             # メタデータ作成(各VCで定義)
             if params.voiceChangerType == "RVC":
@@ -188,6 +193,7 @@ class VoiceChangerManager(ServerDeviceCallbacks):
         data["modelSlots"] = self.modelSlotManager.getAllSlotInfo(reload=True)
         data["sampleModels"] = getSampleInfos(self.params.sample_mode)
         data["python"] = sys.version
+        data["voiceChangerParams"] = self.params
 
         data["status"] = "OK"
 
@@ -214,11 +220,18 @@ class VoiceChangerManager(ServerDeviceCallbacks):
             return
         elif slotInfo.voiceChangerType == "RVC":
             logger.info("................RVC")
-            from voice_changer.RVC.RVC import RVC
+            # from voice_changer.RVC.RVC import RVC
 
-            self.voiceChangerModel = RVC(self.params, slotInfo)
+            # self.voiceChangerModel = RVC(self.params, slotInfo)
-            self.voiceChanger = VoiceChanger(self.params)
+            # self.voiceChanger = VoiceChanger(self.params)
+            # self.voiceChanger.setModel(self.voiceChangerModel)
+
+            from voice_changer.RVC.RVCr2 import RVCr2
+
+            self.voiceChangerModel = RVCr2(self.params, slotInfo)
+            self.voiceChanger = VoiceChangerV2(self.params)
             self.voiceChanger.setModel(self.voiceChangerModel)
+
         elif slotInfo.voiceChangerType == "MMVCv13":
             logger.info("................MMVCv13")
             from voice_changer.MMVCv13.MMVCv13 import MMVCv13
@@ -260,10 +273,16 @@ class VoiceChangerManager(ServerDeviceCallbacks):
                 del self.voiceChangerModel
             return
 
-    def update_settings(self, key: str, val: str | int | float):
+    def update_settings(self, key: str, val: str | int | float | bool):
         self.store_setting(key, val)
 
-        if key in self.settings.intData:
+        if key in self.settings.boolData:
+            if val == "true":
+                newVal = True
+            elif val == "false":
+                newVal = False
+            setattr(self.settings, key, newVal)
+        elif key in self.settings.intData:
             newVal = int(val)
             if key == "modelSlotIndex":
                 newVal = newVal % 1000
@@ -283,6 +302,9 @@ class VoiceChangerManager(ServerDeviceCallbacks):
         return self.get_info()
 
     def changeVoice(self, receivedData: AudioInOut):
+        if self.settings.passThrough is True:  # パススルー
+            return receivedData, []
+
         if hasattr(self, "voiceChanger") is True:
             return self.voiceChanger.on_request(receivedData)
         else:
@@ -299,8 +321,8 @@ class VoiceChangerManager(ServerDeviceCallbacks):
         req.files = [MergeElement(**f) for f in req.files]
         slot = len(self.modelSlotManager.getAllSlotInfo()) - 1
         if req.voiceChangerType == "RVC":
-            merged = RVCModelMerger.merge_models(req, slot)
+            merged = RVCModelMerger.merge_models(self.params, req, slot)
-            loadParam = LoadModelParams(voiceChangerType="RVC", slot=slot, isSampleMode=False, sampleId="", files=[LoadModelParamFile(name=os.path.basename(merged), kind="rvcModel", dir=f"{slot}")], params={})
+            loadParam = LoadModelParams(voiceChangerType="RVC", slot=slot, isSampleMode=False, sampleId="", files=[LoadModelParamFile(name=os.path.basename(merged), kind="rvcModel", dir="")], params={})
             self.loadModel(loadParam)
         return self.get_info()
 

server/voice_changer/VoiceChangerParamsManager.py

@@ -0,0 +1,17 @@
+from voice_changer.utils.VoiceChangerParams import VoiceChangerParams
+
+
+class VoiceChangerParamsManager:
+    _instance = None
+
+    def __init__(self):
+        self.params = None
+
+    @classmethod
+    def get_instance(cls):
+        if cls._instance is None:
+            cls._instance = cls()
+        return cls._instance
+    
+    def setParams(self, params: VoiceChangerParams):
+        self.params = params

server/voice_changer/VoiceChangerV2.py

@@ -6,7 +6,7 @@
 
 - 適用VoiceChangerModel
 ・DiffusionSVC
-
+・RVC
 '''
 
 from typing import Any, Union
@@ -208,12 +208,13 @@ class VoiceChangerV2(VoiceChangerIF):
                 block_frame = receivedData.shape[0]
                 crossfade_frame = min(self.settings.crossFadeOverlapSize, block_frame)
                 self._generate_strength(crossfade_frame)
-                # data = self.voiceChanger.generate_input(newData, block_frame, crossfade_frame, sola_search_frame)
+            
                 audio = self.voiceChanger.inference(
                     receivedData,
                     crossfade_frame=crossfade_frame,
                     sola_search_frame=sola_search_frame
                 )
+
                 if hasattr(self, "sola_buffer") is True:
                     np.set_printoptions(threshold=10000)
                     audio_offset = -1 * (sola_search_frame + crossfade_frame + block_frame)

server/voice_changer/utils/ModelMerger.py

@@ -5,7 +5,7 @@ from dataclasses import dataclass
 
 @dataclass
 class MergeElement:
-    filename: str
+    slotIndex: int
     strength: int
 
 

server/voice_changer/utils/Timer.py

@@ -1,15 +1,43 @@
 import time
+import inspect
 
 
 class Timer(object):
-    def __init__(self, title: str):
+    storedSecs = {}  # Class variable
+
+    def __init__(self, title: str, enalbe: bool = True):
         self.title = title
+        self.enable = enalbe
+        self.secs = 0
+        self.msecs = 0
+        self.avrSecs = 0
+
+        if self.enable is False:
+            return
+
+        self.maxStores = 10
+
+        current_frame = inspect.currentframe()
+        caller_frame = inspect.getouterframes(current_frame, 2)
+        frame = caller_frame[1]
+        filename = frame.filename
+        line_number = frame.lineno
+        self.key = f"{title}_{filename}_{line_number}"
+        if self.key not in self.storedSecs:
+            self.storedSecs[self.key] = []
 
     def __enter__(self):
+        if self.enable is False:
+            return
         self.start = time.time()
         return self
 
     def __exit__(self, *_):
+        if self.enable is False:
+            return
         self.end = time.time()
         self.secs = self.end - self.start
         self.msecs = self.secs * 1000  # millisecs
+        self.storedSecs[self.key].append(self.secs)
+        self.storedSecs[self.key] = self.storedSecs[self.key][-self.maxStores:]
+        self.avrSecs = sum(self.storedSecs[self.key]) / len(self.storedSecs[self.key])

signatures/version1/cla.json

@@ -0,0 +1,20 @@
+{
+  "signedContributors": [
+    {
+      "name": "w-okada",
+      "id": 48346627,
+      "comment_id": 1667673774,
+      "created_at": "2023-08-07T11:21:42Z",
+      "repoId": 527419347,
+      "pullRequestNo": 661
+    },
+    {
+      "name": "w-okada",
+      "id": 48346627,
+      "comment_id": 1667674735,
+      "created_at": "2023-08-07T11:22:28Z",
+      "repoId": 527419347,
+      "pullRequestNo": 661
+    }
+  ]
+}

tutorials/tutorial_rvc_en_latest.md

@@ -44,6 +44,8 @@ If you have the old version, be sure to unzip it into a separate folder.
 
 When connecting remotely, please use `.bat` file (win) and `.command` file (mac) where http is replaced with https.
 
+Access with Browser (currently only chrome is supported), then you can see gui.
+
 ### Console
 
 When you run a .bat file (Windows) or .command file (Mac), a screen like the following will be displayed and various data will be downloaded from the Internet at the initial start-up. Depending on your environment, it may take 1-2 minutes in many cases.

tutorials/tutorial_rvc_ja_latest.md

@@ -44,6 +44,8 @@
 
 リモートから接続する場合は、`.bat`ファイル(win)、`.command`ファイル(mac)の http が https に置き換わっているものを使用してください。
 
+ブラウザ(Chrome のみサポート)でアクセスすると画面が表示されます。
+
 ### コンソール表示
 
 `.bat`ファイル(win)や`.command`ファイル(mac)を実行すると、次のような画面が表示され、初回起動時には各種データをインターネットからダウンロードします。