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

Merge pull request #127 from w-okada/v.1.3.2

Browse Source 
V.1.3.2
This commit is contained in:
w-okada 2023-01-12 21:44:51 +09:00 committed by GitHub
commit 3afaeea4ac
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
11 changed files with 137 additions and 45 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

28
client/demo/src/101_server_setting.tsx
View File

@ -1,4 +1,4 @@
import { OnnxExecutionProvider, Protocol, Framework, fileSelector } from "@dannadori/voice-changer-client-js"
import { OnnxExecutionProvider, Framework, fileSelector } from "@dannadori/voice-changer-client-js"
import React from "react"
import { useMemo } from "react"
import { ClientState } from "./hooks/useClient"
@ -126,29 +126,6 @@ export const useServerSettingArea = (props: UseServerSettingProps): ServerSettin
props.clientState.serverSetting.isUploading,
props.clientState.serverSetting.uploadProgress])
const protocolRow = useMemo(() => {
const onProtocolChanged = async (val: Protocol) => {
props.clientState.clientSetting.setProtocol(val)
}
return (
<div className="body-row split-3-7 left-padding-1 guided">
<div className="body-item-title left-padding-1">Protocol</div>
<div className="body-select-container">
<select className="body-select" value={props.clientState.clientSetting.setting.protocol} onChange={(e) => {
onProtocolChanged(e.target.value as
Protocol)
}}>
{
Object.values(Protocol).map(x => {
return <option key={x} value={x}>{x}</option>
})
}
</select>
</div>
</div>
)
}, [props.clientState.clientSetting.setting.protocol, props.clientState.clientSetting.setProtocol])
const frameworkRow = useMemo(() => {
const onFrameworkChanged = async (val: Framework) => {
props.clientState.serverSetting.setFramework(val)
@ -209,10 +186,9 @@ export const useServerSettingArea = (props: UseServerSettingProps): ServerSettin
{uploadeModelRow}
{frameworkRow}
{onnxExecutionProviderRow}
{protocolRow}
</>
)
}, [uploadeModelRow, frameworkRow, onnxExecutionProviderRow, protocolRow])
}, [uploadeModelRow, frameworkRow, onnxExecutionProviderRow])
return {

23
client/demo/src/102_device_setting.tsx
View File

@ -1,6 +1,6 @@
import { fileSelectorAsDataURL, createDummyMediaStream, SampleRate } from "@dannadori/voice-changer-client-js"
import React, { useEffect, useMemo, useState } from "react"
import { AUDIO_ELEMENT_FOR_PLAY_RESULT, AUDIO_ELEMENT_FOR_TEST_CONVERTED, AUDIO_ELEMENT_FOR_TEST_ORIGINAL } from "./const"
import { AUDIO_ELEMENT_FOR_PLAY_RESULT, AUDIO_ELEMENT_FOR_TEST_CONVERTED, AUDIO_ELEMENT_FOR_TEST_CONVERTED_ECHOBACK, AUDIO_ELEMENT_FOR_TEST_ORIGINAL } from "./const"
import { ClientState } from "./hooks/useClient";
@ -44,6 +44,7 @@ export const useDeviceSetting = (audioContext: AudioContext | null, props: UseDe
const [audioInputForGUI, setAudioInputForGUI] = useState<string>("none")
const [audioOutputForGUI, setAudioOutputForGUI] = useState<string>("none")
const [fileInputEchoback, setFileInputEchoback] = useState<boolean>()//最初のmuteが有効になるように。undefined
useEffect(() => {
@ -114,6 +115,12 @@ export const useDeviceSetting = (audioContext: AudioContext | null, props: UseDe
const dst = audioContext!.createMediaStreamDestination()
src.connect(dst)
props.clientState.clientSetting.setAudioInput(dst.stream)
const audio_echo = document.getElementById(AUDIO_ELEMENT_FOR_TEST_CONVERTED_ECHOBACK) as HTMLAudioElement
audio_echo.srcObject = dst.stream
audio_echo.play()
setFileInputEchoback(false)
// original stream to play.
const audio_org = document.getElementById(AUDIO_ELEMENT_FOR_TEST_ORIGINAL) as HTMLAudioElement
audio_org.src = url
@ -135,14 +142,16 @@ export const useDeviceSetting = (audioContext: AudioContext | null, props: UseDe
</div>
<div>
cnv:<audio id={AUDIO_ELEMENT_FOR_TEST_CONVERTED} controls></audio>
<audio id={AUDIO_ELEMENT_FOR_TEST_CONVERTED_ECHOBACK} controls hidden></audio>
</div>
</div>
<div className="body-button-container">
<div className="body-button" onClick={onFileLoadClicked}>load</div>
<input type="checkbox" checked={fileInputEchoback} onChange={(e) => { setFileInputEchoback(e.target.checked) }} /> echoback
</div>
</div>
)
}, [audioInputForGUI, props.clientState.clientSetting.setAudioInput])
}, [audioInputForGUI, props.clientState.clientSetting.setAudioInput, fileInputEchoback])
@ -166,7 +175,7 @@ export const useDeviceSetting = (audioContext: AudioContext | null, props: UseDe
useEffect(() => {
if (audioOutputForGUI == "none") return
[AUDIO_ELEMENT_FOR_PLAY_RESULT, AUDIO_ELEMENT_FOR_TEST_ORIGINAL].forEach(x => {
[AUDIO_ELEMENT_FOR_PLAY_RESULT, AUDIO_ELEMENT_FOR_TEST_ORIGINAL, AUDIO_ELEMENT_FOR_TEST_CONVERTED_ECHOBACK].forEach(x => {
const audio = document.getElementById(x) as HTMLAudioElement
if (audio) {
// @ts-ignore
@ -175,6 +184,14 @@ export const useDeviceSetting = (audioContext: AudioContext | null, props: UseDe
})
}, [audioOutputForGUI])
useEffect(() => {
[AUDIO_ELEMENT_FOR_TEST_CONVERTED_ECHOBACK].forEach(x => {
const audio = document.getElementById(x) as HTMLAudioElement
if (audio) {
audio.volume = fileInputEchoback ? 1 : 0
}
})
}, [fileInputEchoback])
const deviceSetting = useMemo(() => {
return (

3
client/demo/src/104_convert_setting.tsx
View File

@ -1,5 +1,4 @@
import { BufferSize } from "@dannadori/voice-changer-client-js"
import React, { useMemo, useState } from "react"
import React, { useMemo } from "react"
import { ClientState } from "./hooks/useClient"
export type UseConvertSettingProps = {

44
client/demo/src/105_advanced_setting.tsx
View File

@ -1,4 +1,4 @@
import { BufferSize, SampleRate, VoiceChangerMode } from "@dannadori/voice-changer-client-js"
import { BufferSize, Protocol, SampleRate, VoiceChangerMode } from "@dannadori/voice-changer-client-js"
import React, { useMemo, useState } from "react"
import { ClientState } from "./hooks/useClient"
@ -31,6 +31,30 @@ export const useAdvancedSetting = (props: UseAdvancedSettingProps): AdvancedSett
)
}, [props.clientState.clientSetting.setting.mmvcServerUrl, props.clientState.clientSetting.setServerUrl])
const protocolRow = useMemo(() => {
const onProtocolChanged = async (val: Protocol) => {
props.clientState.clientSetting.setProtocol(val)
}
return (
<div className="body-row split-3-7 left-padding-1 guided">
<div className="body-item-title left-padding-1">Protocol</div>
<div className="body-select-container">
<select className="body-select" value={props.clientState.clientSetting.setting.protocol} onChange={(e) => {
onProtocolChanged(e.target.value as
Protocol)
}}>
{
Object.values(Protocol).map(x => {
return <option key={x} value={x}>{x}</option>
})
}
</select>
</div>
</div>
)
}, [props.clientState.clientSetting.setting.protocol, props.clientState.clientSetting.setProtocol])
const sampleRateRow = useMemo(() => {
return (
<div className="body-row split-3-7 left-padding-1 guided">
@ -84,6 +108,20 @@ export const useAdvancedSetting = (props: UseAdvancedSettingProps): AdvancedSett
)
}, [props.clientState.serverSetting.setting.convertChunkNum, props.clientState.serverSetting.setConvertChunkNum])
const minConvertSizeRow = useMemo(() => {
return (
<div className="body-row split-3-7 left-padding-1 guided">
<div className="body-item-title left-padding-1">Min Convert Size(byte)</div>
<div className="body-input-container">
<input type="number" min={0} max={8196} step={8196} value={props.clientState.serverSetting.setting.minConvertSize} onChange={(e) => {
props.clientState.serverSetting.setMinConvertSize(Number(e.target.value))
}} />
</div>
</div>
)
}, [props.clientState.serverSetting.setting.minConvertSize, props.clientState.serverSetting.setMinConvertSize])
const crossFadeOverlapRateRow = useMemo(() => {
return (
<div className="body-row split-3-7 left-padding-1 guided">
@ -209,12 +247,14 @@ export const useAdvancedSetting = (props: UseAdvancedSettingProps): AdvancedSett
<>
<div className="body-row divider"></div>
{mmvcServerUrlRow}
{protocolRow}
<div className="body-row divider"></div>
{sampleRateRow}
{bufferSizeRow}
<div className="body-row divider"></div>
{convertChunkNumRow}
{minConvertSizeRow}
{crossFadeOverlapRateRow}
{crossFadeOffsetRateRow}
{crossFadeEndRateRow}
@ -226,7 +266,7 @@ export const useAdvancedSetting = (props: UseAdvancedSettingProps): AdvancedSett
<div className="body-row divider"></div>
</>
)
}, [showAdvancedSetting, mmvcServerUrlRow, sampleRateRow, bufferSizeRow, convertChunkNumRow, crossFadeOverlapRateRow, crossFadeOffsetRateRow, crossFadeEndRateRow, vfForceDisableRow, voiceChangeModeRow, workletSettingRow])
}, [showAdvancedSetting, mmvcServerUrlRow, protocolRow, sampleRateRow, bufferSizeRow, convertChunkNumRow, minConvertSizeRow, crossFadeOverlapRateRow, crossFadeOffsetRateRow, crossFadeEndRateRow, vfForceDisableRow, voiceChangeModeRow, workletSettingRow])
const advancedSetting = useMemo(() => {

1
client/demo/src/const.ts
View File

@ -3,3 +3,4 @@ export const CHROME_EXTENSION = false
export const AUDIO_ELEMENT_FOR_PLAY_RESULT = "audio-result"
export const AUDIO_ELEMENT_FOR_TEST_ORIGINAL = "audio-test-original"
export const AUDIO_ELEMENT_FOR_TEST_CONVERTED = "audio-test-converted"
export const AUDIO_ELEMENT_FOR_TEST_CONVERTED_ECHOBACK = "audio-test-converted-echoback"

16
client/demo/src/hooks/useClientSetting.ts
View File

@ -150,19 +150,23 @@ export const useClientSetting = (props: UseClientSettingProps): ClientSettingSta
}
}, [props.voiceChangerClient])
//////////////
// Colab対応
// デフォルト設定
/////////////
useEffect(() => {
const params = new URLSearchParams(location.search);
const colab = params.get("colab")
if (colab == "true") {
settingRef.current.protocol = "rest"
settingRef.current.inputChunkNum = 64
_setSetting({ ...settingRef.current })
setProtocol("rest")
setInputChunkNum(64)
} else {
setProtocol("sio")
setInputChunkNum(32)
}
}, [])
}, [props.voiceChangerClient])
return {

28
client/demo/src/hooks/useServerSetting.ts
View File

@ -1,5 +1,5 @@
import { DefaultVoiceChangerServerSetting, Framework, OnnxExecutionProvider, ServerInfo, ServerSettingKey, VoiceChangerClient, VoiceChangerServerSetting, } from "@dannadori/voice-changer-client-js"
import { useState, useMemo, useRef, } from "react"
import { useState, useMemo, useRef, useEffect } from "react"
export type FileUploadSetting = {
@ -25,6 +25,7 @@ export type ServerSettingState = {
setSrcId: (num: number) => Promise<boolean>;
setDstId: (num: number) => Promise<boolean>;
setConvertChunkNum: (num: number) => Promise<boolean>;
setMinConvertSize: (num: number) => Promise<boolean>
setGpu: (num: number) => Promise<boolean>;
setCrossFadeOffsetRate: (num: number) => Promise<boolean>;
setCrossFadeEndRate: (num: number) => Promise<boolean>;
@ -56,6 +57,7 @@ export const useServerSetting = (props: UseServerSettingProps): ServerSettingSta
_setSetting({
...settingRef.current,
convertChunkNum: res.convertChunkNum,
minConvertSize: res.minConvertSize,
srcId: res.srcId,
dstId: res.dstId,
gpu: res.gpu,
@ -103,6 +105,13 @@ export const useServerSetting = (props: UseServerSettingProps): ServerSettingSta
}
}, [props.voiceChangerClient])
const setMinConvertSize = useMemo(() => {
return async (num: number) => {
return await _set_and_store(ServerSettingKey.minConvertSize, "" + num)
}
}, [props.voiceChangerClient])
const setGpu = useMemo(() => {
return async (num: number) => {
return await _set_and_store(ServerSettingKey.gpu, "" + num)
@ -125,6 +134,8 @@ export const useServerSetting = (props: UseServerSettingProps): ServerSettingSta
}
}, [props.voiceChangerClient])
//////////////
// 操作
/////////////
@ -195,6 +206,20 @@ export const useServerSetting = (props: UseServerSettingProps): ServerSettingSta
}
}, [props.voiceChangerClient])
//////////////
// デフォルト設定
/////////////
useEffect(() => {
const params = new URLSearchParams(location.search);
const colab = params.get("colab")
if (colab == "true") {
} else {
}
}, [props.voiceChangerClient])
return {
setting,
serverInfo,
@ -204,6 +229,7 @@ export const useServerSetting = (props: UseServerSettingProps): ServerSettingSta
setSrcId,
setDstId,
setConvertChunkNum,
setMinConvertSize,
setGpu,
setCrossFadeOffsetRate,
setCrossFadeEndRate,

24
client/demo/src/hooks/useWorkletSetting.ts
View File

@ -1,5 +1,5 @@
import { DefaultWorkletSetting, VoiceChangerClient, WorkletSetting } from "@dannadori/voice-changer-client-js"
import { useState, useMemo } from "react"
import { useState, useMemo, useEffect } from "react"
export type UseWorkletSettingProps = {
voiceChangerClient: VoiceChangerClient | null
@ -22,6 +22,28 @@ export const useWorkletSetting = (props: UseWorkletSettingProps): WorkletSetting
}
}, [props.voiceChangerClient])
//////////////
// デフォルト設定
/////////////
useEffect(() => {
const params = new URLSearchParams(location.search);
const colab = params.get("colab")
if (colab == "true") {
setSetting({
numTrancateTreshold: 300,
volTrancateThreshold: 0.0005,
volTrancateLength: 32,
})
} else {
setSetting({
numTrancateTreshold: 150,
volTrancateThreshold: 0.0005,
volTrancateLength: 32,
})
}
}, [props.voiceChangerClient])
return {
setting,
setSetting

5
client/lib/src/const.ts
View File

@ -7,6 +7,8 @@
// types
export type VoiceChangerServerSetting = {
convertChunkNum: number, // VITSに入力する変換サイズ。(入力データの2倍以上の大きさで指定。それより小さいものが指定された場合は、サーバ側で自動的に入力の2倍のサイズが設定される。)
minConvertSize: number, // この値より小さい場合にこの値に揃える。
srcId: number,
dstId: number,
gpu: number,
@ -50,6 +52,7 @@ export type ServerInfo = {
pyTorchModelFile: string,
onnxModelFile: string,
convertChunkNum: number,
minConvertSize: number,
crossFadeOffsetRate: number,
crossFadeEndRate: number,
crossFadeOverlapRate: number,
@ -111,6 +114,7 @@ export const ServerSettingKey = {
"srcId": "srcId",
"dstId": "dstId",
"convertChunkNum": "convertChunkNum",
"minConvertSize": "minConvertSize",
"gpu": "gpu",
"crossFadeOffsetRate": "crossFadeOffsetRate",
"crossFadeEndRate": "crossFadeEndRate",
@ -123,6 +127,7 @@ export type ServerSettingKey = typeof ServerSettingKey[keyof typeof ServerSettin
// Defaults
export const DefaultVoiceChangerServerSetting: VoiceChangerServerSetting = {
convertChunkNum: 32, //（★１）
minConvertSize: 0,
srcId: 107,
dstId: 100,
gpu: 0,

8
server/voice_changer/VoiceChanger.py
View File

@ -26,12 +26,13 @@ class VocieChangerSettings():
crossFadeEndRate:float = 0.9
crossFadeOverlapRate:float = 0.9
convertChunkNum:int = 32
minConvertSize:int = 0
framework:str = "PyTorch" # PyTorch or ONNX
pyTorchModelFile:str = ""
onnxModelFile:str = ""
configFile:str = ""
# ↓mutableな物だけ列挙
intData = ["gpu","srcId", "dstId", "convertChunkNum"]
intData = ["gpu","srcId", "dstId", "convertChunkNum", "minConvertSize"]
floatData = [ "crossFadeOffsetRate", "crossFadeEndRate", "crossFadeOverlapRate"]
strData = ["framework"]
@ -298,8 +299,9 @@ class VoiceChanger():
if unpackedData.shape[0]*(1 + self.settings.crossFadeOverlapRate) + 1024 > convertSize:
convertSize = int(unpackedData.shape[0]*(1 + self.settings.crossFadeOverlapRate)) + 1024
# print("convert Size", unpackedData.shape[0], unpackedData.shape[0]*(1 + self.settings.crossFadeOverlapRate), convertSize)
if convertSize < self.settings.minConvertSize:
convertSize = self.settings.minConvertSize
# print("convert Size", unpackedData.shape[0], unpackedData.shape[0]*(1 + self.settings.crossFadeOverlapRate), convertSize, self.settings.minConvertSize)
self._generate_strength(unpackedData)
data = self._generate_input(unpackedData, convertSize)