1berry/wokada-voice-changer
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

update information when upload model

Browse Source
This commit is contained in:
wataru 2023-01-28 15:56:56 +09:00
parent eb846bedcc
commit 9a473a07be
8 changed files with 6728 additions and 3172 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
client/demo/dist/index.js vendored
View File

File diff suppressed because one or more lines are too long

7048
client/demo/package-lock.json generated
View File

File diff suppressed because it is too large Load Diff

4
client/demo/package.json
View File

@ -38,7 +38,7 @@
"postcss-loader": "^7.0.2",
"postcss-nested": "^6.0.0",
"prettier": "^2.8.3",
"rimraf": "^4.1.1",
"rimraf": "^4.1.2",
"style-loader": "^3.3.1",
"ts-loader": "^9.4.2",
"tsconfig-paths": "^4.1.2",
@ -48,7 +48,7 @@
"webpack-dev-server": "^4.11.1"
},
"dependencies": {
"@dannadori/voice-changer-client-js": "^1.0.20",
"@dannadori/voice-changer-client-js": "^1.0.28",
"react": "^18.2.0",
"react-dom": "^18.2.0"
}

2718
client/lib/package-lock.json generated
View File

File diff suppressed because it is too large Load Diff

8
client/lib/package.json
View File

@ -1,6 +1,6 @@
{
"name": "@dannadori/voice-changer-client-js",
"version": "1.0.20",
"version": "1.0.28",
"description": "",
"main": "dist/index.js",
"directories": {
@ -26,17 +26,17 @@
"devDependencies": {
"@types/audioworklet": "^0.0.36",
"@types/node": "^18.11.18",
"@types/react": "18.0.26",
"@types/react": "18.0.27",
"@types/react-dom": "18.0.10",
"eslint": "^8.32.0",
"eslint-config-prettier": "^8.6.0",
"eslint-plugin-prettier": "^4.2.1",
"eslint-plugin-react": "^7.32.0",
"eslint-plugin-react": "^7.32.1",
"eslint-webpack-plugin": "^3.2.0",
"npm-run-all": "^4.1.5",
"prettier": "^2.8.3",
"raw-loader": "^4.0.2",
"rimraf": "^4.0.7",
"rimraf": "^4.1.2",
"ts-loader": "^9.4.2",
"typescript": "^4.9.4",
"webpack": "^5.75.0",

2
client/lib/src/const.ts
View File

@ -135,7 +135,7 @@ export const DefaultVoiceChangerServerSetting: VoiceChangerServerSetting = {
crossFadeOffsetRate: 0.1,
crossFadeEndRate: 0.9,
crossFadeOverlapRate: 0.5,
framework: "PyTorch",
framework: "ONNX",
onnxExecutionProvider: "CPUExecutionProvider"
}

5
client/lib/src/hooks/useServerSetting.ts
View File

@ -179,10 +179,10 @@ export const useServerSetting = (props: UseServerSettingProps): ServerSettingSta
console.log(progress, end)
})
const serverInfo = await props.voiceChangerClient.loadModel(fileUploadSetting.configFile, fileUploadSetting.pyTorchModel, fileUploadSetting.onnxModel)
console.log(serverInfo)
await props.voiceChangerClient.loadModel(fileUploadSetting.configFile, fileUploadSetting.pyTorchModel, fileUploadSetting.onnxModel)
setUploadProgress(0)
setIsUploading(false)
reloadServerInfo()
}
}, [fileUploadSetting, props.voiceChangerClient])
@ -203,7 +203,6 @@ export const useServerSetting = (props: UseServerSettingProps): ServerSettingSta
framework: res.framework,
onnxExecutionProvider: (!!res.onnxExecutionProvider && res.onnxExecutionProvider.length > 0) ? res.onnxExecutionProvider[0] as OnnxExecutionProvider : DefaultVoiceChangerServerSetting.onnxExecutionProvider
})
}
}, [props.voiceChangerClient])

21
server/voice_changer/VoiceChanger.py
View File

@ -1,6 +1,7 @@
from const import ERROR_NO_ONNX_SESSION
import torch
import os, traceback
import os
import traceback
import numpy as np
from dataclasses import dataclass, asdict
@ -13,6 +14,7 @@ from voice_changer.TrainerFunctions import TextAudioSpeakerCollate, spectrogram_
providers = ['OpenVINOExecutionProvider', "CUDAExecutionProvider", "DmlExecutionProvider", "CPUExecutionProvider"]
@dataclass
class VocieChangerSettings():
gpu: int = 0
@ -23,7 +25,7 @@ class VocieChangerSettings():
crossFadeOverlapRate: float = 0.9
convertChunkNum: int = 32
minConvertSize: int = 0
framework:str = "PyTorch" # PyTorch or ONNX
framework: str = "ONNX" # PyTorch or ONNX
pyTorchModelFile: str = ""
onnxModelFile: str = ""
configFile: str = ""
@ -32,6 +34,7 @@ class VocieChangerSettings():
floatData = ["crossFadeOffsetRate", "crossFadeEndRate", "crossFadeOverlapRate"]
strData = ["framework"]
class VoiceChanger():
def __init__(self, config: str):
@ -136,7 +139,6 @@ class VoiceChanger():
return self.get_info()
def _generate_strength(self, unpackedData):
if self.unpackedData_length != unpackedData.shape[0] or self.currentCrossFadeOffsetRate != self.settings.crossFadeOffsetRate or self.currentCrossFadeEndRate != self.settings.crossFadeEndRate or self.currentCrossFadeOverlapRate != self.settings.crossFadeOverlapRate:
@ -191,7 +193,6 @@ class VoiceChanger():
data = TextAudioSpeakerCollate()([data])
return data
def _onnx_inference(self, data, inputSize):
if hasattr(self, "onnx_session") == False or self.onnx_session == None:
print("[Voice Changer] No ONNX session.")
@ -234,7 +235,8 @@ class VoiceChanger():
with torch.no_grad():
x, x_lengths, spec, spec_lengths, y, y_lengths, sid_src = [x.cpu() for x in data]
sid_tgt1 = torch.LongTensor([self.settings.dstId]).cpu()
audio1 = (self.net_g.cpu().voice_conversion(spec, spec_lengths, sid_src=sid_src, sid_tgt=sid_tgt1)[0, 0].data * self.hps.data.max_wav_value)
audio1 = (self.net_g.cpu().voice_conversion(spec, spec_lengths, sid_src=sid_src,
sid_tgt=sid_tgt1)[0, 0].data * self.hps.data.max_wav_value)
if self.prev_strength.device != torch.device('cpu'):
print(f"prev_strength move from {self.prev_strength.device} to cpu")
@ -265,7 +267,8 @@ class VoiceChanger():
with torch.no_grad():
x, x_lengths, spec, spec_lengths, y, y_lengths, sid_src = [x.cuda(self.settings.gpu) for x in data]
sid_tgt1 = torch.LongTensor([self.settings.dstId]).cuda(self.settings.gpu)
audio1 = self.net_g.cuda(self.settings.gpu).voice_conversion(spec, spec_lengths, sid_src=sid_src, sid_tgt=sid_tgt1)[0, 0].data * self.hps.data.max_wav_value
audio1 = self.net_g.cuda(self.settings.gpu).voice_conversion(spec, spec_lengths, sid_src=sid_src,
sid_tgt=sid_tgt1)[0, 0].data * self.hps.data.max_wav_value
if self.prev_strength.device != torch.device('cuda', self.settings.gpu):
print(f"prev_strength move from {self.prev_strength.device} to gpu{self.settings.gpu}")
@ -274,8 +277,6 @@ class VoiceChanger():
print(f"cur_strength move from {self.cur_strength.device} to gpu{self.settings.gpu}")
self.cur_strength = self.cur_strength.cuda(self.settings.gpu)
if hasattr(self, 'prev_audio1') == True and self.prev_audio1.device == torch.device('cuda', self.settings.gpu):
overlapSize = int(inputSize * self.settings.crossFadeOverlapRate)
prev_overlap = self.prev_audio1[-1 * overlapSize:]
@ -295,7 +296,6 @@ class VoiceChanger():
result = result.cpu().float().numpy()
return result
def on_request(self, unpackedData: any):
convertSize = self.settings.convertChunkNum * 128 # 128sample/1chunk
@ -308,14 +308,12 @@ class VoiceChanger():
self._generate_strength(unpackedData)
data = self._generate_input(unpackedData, convertSize)
try:
if self.settings.framework == "ONNX":
result = self._onnx_inference(data, unpackedData.shape[0])
else:
result = self._pyTorch_inference(data, unpackedData.shape[0])
except Exception as e:
print("VC PROCESSING!!!! EXCEPTION!!!", e)
print(traceback.format_exc())
@ -328,4 +326,3 @@ class VoiceChanger():
result = result.astype(np.int16)
# print("on_request result size:",result.shape)
return result