PublicMirrors/voice-changer
1
0
Fork 0
mirror of https://github.com/w-okada/voice-changer.git synced 2025-01-23 13:35:12 +03:00
Code Issues Actions Packages Projects Releases Wiki Activity

uploading status

Browse Source
This commit is contained in:
wataru 2023-01-08 20:28:57 +09:00
parent 8082783644
commit 276f086823
11 changed files with 164 additions and 2899 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

2744
client/demo/dist/index.js.LICENSE.txt vendored
View File

File diff suppressed because it is too large Load Diff

87
client/demo/src/100_options_microphone.tsx
View File

@ -14,13 +14,19 @@ import { ServerSettingKey } from "@dannadori/voice-changer-client-js";
export const useMicrophoneOptions = () => {
const [audioContext, setAudioContext] = useState<AudioContext | null>(null)
const [loadModelFunc, setLoadModelFunc] = useState<() => Promise<void>>()
const [uploadProgress, setUploadProgress] = useState<number>(0)
const [isUploading, setIsUploading] = useState<boolean>(false)
const clientState = useClient({
audioContext: audioContext,
audioOutputElementId: AUDIO_ELEMENT_FOR_PLAY_RESULT
})
const serverSetting = useServerSetting({
clientState
clientState,
loadModelFunc,
uploadProgress: uploadProgress,
isUploading: isUploading
})
const deviceSetting = useDeviceSetting(audioContext)
const speakerSetting = useSpeakerSetting()
@ -34,7 +40,8 @@ export const useMicrophoneOptions = () => {
getInfo: clientState.getInfo,
volume: clientState.volume,
bufferingTime: clientState.bufferingTime,
responseTime: clientState.responseTime
responseTime: clientState.responseTime,
})
useEffect(() => {
@ -51,70 +58,94 @@ export const useMicrophoneOptions = () => {
// 101 ServerSetting
//// サーバ変更
useEffect(() => {
if (!clientState.clientInitialized) return
clientState.setServerUrl(serverSetting.mmvcServerUrl)
}, [clientState.clientInitialized, serverSetting.mmvcServerUrl])
}, [serverSetting.mmvcServerUrl])
//// プロトコル変更
useEffect(() => {
if (!clientState.clientInitialized) return
clientState.setProtocol(serverSetting.protocol)
}, [clientState.clientInitialized, serverSetting.protocol])
}, [serverSetting.protocol])
//// フレームワーク変更
useEffect(() => {
if (!clientState.clientInitialized) return
clientState.updateSettings(ServerSettingKey.framework, serverSetting.framework)
}, [clientState.clientInitialized, serverSetting.framework])
}, [serverSetting.framework])
//// OnnxExecutionProvider変更
useEffect(() => {
if (!clientState.clientInitialized) return
clientState.updateSettings(ServerSettingKey.onnxExecutionProvider, serverSetting.onnxExecutionProvider)
}, [clientState.clientInitialized, serverSetting.onnxExecutionProvider])
}, [serverSetting.onnxExecutionProvider])
// 102 DeviceSetting
//// 入力情報の設定
useEffect(() => {
if (!clientState.clientInitialized) return
clientState.changeInput(deviceSetting.audioInput, convertSetting.bufferSize, advancedSetting.vfForceDisabled)
}, [clientState.clientInitialized, deviceSetting.audioInput, convertSetting.bufferSize, advancedSetting.vfForceDisabled])
}, [deviceSetting.audioInput, convertSetting.bufferSize, advancedSetting.vfForceDisabled])
// 103 SpeakerSetting
// 音声変換元、変換先の設定
useEffect(() => {
if (!clientState.clientInitialized) return
clientState.updateSettings(ServerSettingKey.srcId, speakerSetting.srcId)
}, [clientState.clientInitialized, speakerSetting.srcId])
}, [speakerSetting.srcId])
useEffect(() => {
if (!clientState.clientInitialized) return
clientState.updateSettings(ServerSettingKey.dstId, speakerSetting.dstId)
}, [clientState.clientInitialized, speakerSetting.dstId])
}, [speakerSetting.dstId])
// 104 ConvertSetting
useEffect(() => {
if (!clientState.clientInitialized) return
clientState.setInputChunkNum(convertSetting.inputChunkNum)
}, [clientState.clientInitialized, convertSetting.inputChunkNum])
}, [convertSetting.inputChunkNum])
useEffect(() => {
if (!clientState.clientInitialized) return
clientState.updateSettings(ServerSettingKey.convertChunkNum, convertSetting.convertChunkNum)
}, [clientState.clientInitialized, convertSetting.convertChunkNum])
}, [convertSetting.convertChunkNum])
useEffect(() => {
if (!clientState.clientInitialized) return
clientState.updateSettings(ServerSettingKey.gpu, convertSetting.gpu)
}, [clientState.clientInitialized, convertSetting.gpu])
}, [convertSetting.gpu])
useEffect(() => {
if (!clientState.clientInitialized) return
clientState.updateSettings(ServerSettingKey.crossFadeOffsetRate, convertSetting.crossFadeOffsetRate)
}, [clientState.clientInitialized, convertSetting.crossFadeOffsetRate])
}, [convertSetting.crossFadeOffsetRate])
useEffect(() => {
if (!clientState.clientInitialized) return
clientState.updateSettings(ServerSettingKey.crossFadeEndRate, convertSetting.crossFadeEndRate)
}, [clientState.clientInitialized, convertSetting.crossFadeEndRate])
}, [convertSetting.crossFadeEndRate])
// 105 AdvancedSetting
useEffect(() => {
if (!clientState.clientInitialized) return
clientState.setVoiceChangerMode(advancedSetting.voiceChangerMode)
}, [clientState.clientInitialized, advancedSetting.voiceChangerMode])
}, [advancedSetting.voiceChangerMode])
// Model Load
useEffect(() => {
const loadModel = () => {
return async () => {
if (!serverSetting.pyTorchModel && !serverSetting.onnxModel) {
alert("PyTorchモデルとONNXモデルのどちらか一つ以上指定する必要があります。")
return
}
if (!serverSetting.configFile) {
alert("Configファイルを指定する必要があります。")
return
}
setUploadProgress(0)
setIsUploading(true)
const models = [serverSetting.pyTorchModel, serverSetting.onnxModel].filter(x => { return x != null }) as File[]
for (let i = 0; i < models.length; i++) {
const progRate = 1 / models.length
const progOffset = 100 * i * progRate
await clientState.uploadFile(models[i], (progress: number, end: boolean) => {
// console.log(progress * progRate + progOffset, end, progRate,)
setUploadProgress(progress * progRate + progOffset)
})
}
await clientState.uploadFile(serverSetting.configFile, (progress: number, end: boolean) => {
console.log(progress, end)
})
await clientState.loadModel(serverSetting.configFile, serverSetting.pyTorchModel, serverSetting.onnxModel)
setUploadProgress(0)
setIsUploading(false)
}
}
setLoadModelFunc(loadModel)
}, [serverSetting.configFile, serverSetting.pyTorchModel, serverSetting.onnxModel])
const voiceChangerSetting = useMemo(() => {

43
client/demo/src/101_server_setting.tsx
View File

@ -5,6 +5,9 @@ import { ClientState } from "./hooks/useClient"
export type UseServerSettingProps = {
clientState: ClientState
loadModelFunc: (() => Promise<void>) | undefined
uploadProgress: number,
isUploading: boolean
}
export type ServerSettingState = {
@ -16,6 +19,7 @@ export type ServerSettingState = {
framework: string;
onnxExecutionProvider: OnnxExecutionProvider;
protocol: Protocol;
}
export const useServerSetting = (props: UseServerSettingProps): ServerSettingState => {
@ -54,6 +58,9 @@ export const useServerSetting = (props: UseServerSettingProps): ServerSettingSta
}
setPyTorchModel(file)
}
const onPyTorchFileClearClicked = () => {
setPyTorchModel(null)
}
const onConfigFileLoadClicked = async () => {
const file = await fileSelector("")
if (file.name.endsWith(".json") == false) {
@ -62,6 +69,9 @@ export const useServerSetting = (props: UseServerSettingProps): ServerSettingSta
}
setConfigFile(file)
}
const onConfigFileClearClicked = () => {
setConfigFile(null)
}
const onOnnxFileLoadClicked = async () => {
const file = await fileSelector("")
if (file.name.endsWith(".onnx") == false) {
@ -70,31 +80,14 @@ export const useServerSetting = (props: UseServerSettingProps): ServerSettingSta
}
setOnnxModel(file)
}
const onOnnxFileClearClicked = () => {
setOnnxModel(null)
}
const onModelUploadClicked = async () => {
if (!pyTorchModel && !onnxModel) {
alert("PyTorchモデルとONNXモデルのどちらか一つ以上指定する必要があります。")
if (!props.loadModelFunc) {
return
}
if (!configFile) {
alert("Configファイルを指定する必要があります。")
return
}
if (pyTorchModel) {
await props.clientState.uploadFile(pyTorchModel, (progress: number, end: boolean) => {
console.log(progress, end)
})
}
if (onnxModel) {
await props.clientState.uploadFile(onnxModel, (progress: number, end: boolean) => {
console.log(progress, end)
})
}
await props.clientState.uploadFile(configFile, (progress: number, end: boolean) => {
console.log(progress, end)
})
await props.clientState.loadModel(configFile, pyTorchModel, onnxModel)
console.log("loaded")
props.loadModelFunc()
}
return (
@ -115,6 +108,7 @@ export const useServerSetting = (props: UseServerSettingProps): ServerSettingSta
</div>
<div className="body-button-container">
<div className="body-button" onClick={onPyTorchFileLoadClicked}>select</div>
<div className="body-button left-margin-1" onClick={onPyTorchFileClearClicked}>clear</div>
</div>
</div>
<div className="body-row split-3-3-4 left-padding-1 guided">
@ -124,6 +118,7 @@ export const useServerSetting = (props: UseServerSettingProps): ServerSettingSta
</div>
<div className="body-button-container">
<div className="body-button" onClick={onConfigFileLoadClicked}>select</div>
<div className="body-button left-margin-1" onClick={onConfigFileClearClicked}>clear</div>
</div>
</div>
<div className="body-row split-3-3-4 left-padding-1 guided">
@ -133,11 +128,13 @@ export const useServerSetting = (props: UseServerSettingProps): ServerSettingSta
</div>
<div className="body-button-container">
<div className="body-button" onClick={onOnnxFileLoadClicked}>select</div>
<div className="body-button left-margin-1" onClick={onOnnxFileClearClicked}>clear</div>
</div>
</div>
<div className="body-row split-3-3-4 left-padding-1 guided">
<div className="body-item-title left-padding-2"></div>
<div className="body-item-text">
{props.isUploading ? `uploading.... ${props.uploadProgress}%` : ""}
</div>
<div className="body-button-container">
<div className="body-button" onClick={onModelUploadClicked}>upload</div>
@ -145,7 +142,7 @@ export const useServerSetting = (props: UseServerSettingProps): ServerSettingSta
</div>
</>
)
}, [pyTorchModel, configFile, onnxModel, mmvcServerUrl])
}, [pyTorchModel, configFile, onnxModel, props.loadModelFunc, props.isUploading, props.uploadProgress])
const protocolRow = useMemo(() => {
const onProtocolChanged = async (val: Protocol) => {

6
client/demo/src/css/App.css
View File

@ -194,6 +194,12 @@ body {
.left-padding-2 {
padding-left: 2rem;
}
.left-margin-1 {
margin-left: 1rem;
}
.left-margin-2 {
margin-left: 2rem;
}
.highlight {
background-color: rgba(200, 200, 255, 0.3);
}

61
client/demo/src/hooks/useClient.ts
View File

@ -41,6 +41,13 @@ export const useClient = (props: UseClientProps): ClientState => {
const [responseTime, setResponseTime] = useState<number>(0)
const [volume, setVolume] = useState<number>(0)
const initializedResolveRef = useRef<(value: void | PromiseLike<void>) => void>()
const initializedPromise = useMemo(() => {
return new Promise<void>((resolve) => {
initializedResolveRef.current = resolve
})
}, [])
useEffect(() => {
const initialized = async () => {
if (!props.audioContext) {
@ -63,6 +70,7 @@ export const useClient = (props: UseClientProps): ClientState => {
setVolume(vol)
}
})
await voiceChangerClient.isInitialized()
voiceChangerClientRef.current = voiceChangerClient
console.log("[useClient] client initialized")
@ -71,6 +79,7 @@ export const useClient = (props: UseClientProps): ClientState => {
const audio = document.getElementById(props.audioOutputElementId) as HTMLAudioElement
audio.srcObject = voiceChangerClientRef.current.stream
audio.play()
initializedResolveRef.current!()
}
initialized()
}, [props.audioContext])
@ -78,10 +87,7 @@ export const useClient = (props: UseClientProps): ClientState => {
// Client Setting
const setServerUrl = useMemo(() => {
return async (mmvcServerUrl: string) => {
if (!voiceChangerClientRef.current) {
console.log("client is not initialized")
return
}
await initializedPromise
voiceChangerClientRef.current.setServerUrl(mmvcServerUrl, true)
voiceChangerClientRef.current.stop()
}
@ -89,30 +95,21 @@ export const useClient = (props: UseClientProps): ClientState => {
const setProtocol = useMemo(() => {
return async (protocol: Protocol) => {
if (!voiceChangerClientRef.current) {
console.log("client is not initialized")
return
}
await initializedPromise
voiceChangerClientRef.current.setProtocol(protocol)
}
}, [])
const setInputChunkNum = useMemo(() => {
return async (num: number) => {
if (!voiceChangerClientRef.current) {
console.log("client is not initialized")
return
}
await initializedPromise
voiceChangerClientRef.current.setInputChunkNum(num)
}
}, [])
const setVoiceChangerMode = useMemo(() => {
return async (val: VoiceChangerMode) => {
if (!voiceChangerClientRef.current) {
console.log("client is not initialized")
return
}
await initializedPromise
voiceChangerClientRef.current.setVoiceChangerMode(val)
voiceChangerClientRef.current.stop()
}
@ -122,20 +119,14 @@ export const useClient = (props: UseClientProps): ClientState => {
// Client Control
const start = useMemo(() => {
return async (mmvcServerUrl: string) => {
if (!voiceChangerClientRef.current) {
console.log("client is not initialized")
return
}
await initializedPromise
voiceChangerClientRef.current.setServerUrl(mmvcServerUrl, true)
voiceChangerClientRef.current.start()
}
}, [])
const stop = useMemo(() => {
return async () => {
if (!voiceChangerClientRef.current) {
// console.log("client is not initialized")
return
}
await initializedPromise
voiceChangerClientRef.current.stop()
}
}, [])
@ -144,10 +135,8 @@ export const useClient = (props: UseClientProps): ClientState => {
// Device Setting
const changeInput = useMemo(() => {
return async (audioInput: MediaStream | string, bufferSize: BufferSize, vfForceDisable: boolean) => {
if (!voiceChangerClientRef.current || !props.audioContext) {
console.log("[useClient] not initialized", voiceChangerClientRef.current, props.audioContext)
return
}
await initializedPromise
if (!props.audioContext) return
if (!audioInput || audioInput == "none") {
console.log("[useClient] setup!(1)", audioInput)
const ms = createDummyMediaStream(props.audioContext)
@ -165,9 +154,7 @@ export const useClient = (props: UseClientProps): ClientState => {
// Server Setting
const uploadFile = useMemo(() => {
return async (file: File, onprogress: (progress: number, end: boolean) => void) => {
if (!voiceChangerClientRef.current) {
throw "[useClient] Client Not Initialized."
}
await initializedPromise
const num = await voiceChangerClientRef.current.uploadFile(file, onprogress)
const res = await voiceChangerClientRef.current.concatUploadedFile(file, num)
console.log("upload", num, res)
@ -176,9 +163,7 @@ export const useClient = (props: UseClientProps): ClientState => {
const loadModel = useMemo(() => {
return async (configFile: File, pyTorchModelFile: File | null, onnxModelFile: File | null) => {
if (!voiceChangerClientRef.current) {
throw "[useClient] Client Not Initialized."
}
await initializedPromise
await voiceChangerClientRef.current.loadModel(configFile, pyTorchModelFile, onnxModelFile)
console.log("load model")
}
@ -186,9 +171,7 @@ export const useClient = (props: UseClientProps): ClientState => {
const updateSettings = useMemo(() => {
return async (key: ServerSettingKey, val: string | number) => {
if (!voiceChangerClientRef.current) {
throw "[useClient] Client Not Initialized."
}
await initializedPromise
return await voiceChangerClientRef.current.updateServerSettings(key, "" + val)
}
}, [])
@ -196,9 +179,7 @@ export const useClient = (props: UseClientProps): ClientState => {
// Information
const getInfo = useMemo(() => {
return async () => {
if (!voiceChangerClientRef.current) {
throw "[useClient] Client Not Initialized."
}
await initializedPromise
const serverSettings = await voiceChangerClientRef.current.getServerSettings()
const clientSettings = await voiceChangerClientRef.current.getClientSettings()
console.log(serverSettings, clientSettings)

6
client/lib/src/ServerConfigurator.ts
View File

@ -59,7 +59,7 @@ export class ServerConfigurator {
}
const chunkNum = fileChunks.length
console.log("FILE_CHUNKS:", chunkNum, fileChunks)
// console.log("FILE_CHUNKS:", chunkNum, fileChunks)
while (true) {
@ -77,8 +77,8 @@ export class ServerConfigurator {
method: 'POST',
body: formData,
});
fetch(request).then(async (response) => {
console.log(await response.text())
fetch(request).then(async (_response) => {
// console.log(await response.text())
resolve()
})
})

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

@ -122,9 +122,7 @@ export class VoiceChnagerClient {
// create mic stream
if (this.micStream) {
console.log("DESTROY!!!!!!!!!!!!!!!!!!!")
this.micStream.unpipe()
// this.micStream.stop()
this.micStream.destroy()
this.micStream = null
}
@ -176,9 +174,8 @@ export class VoiceChnagerClient {
setServerUrl = (serverUrl: string, openTab: boolean = false) => {
const url = validateUrl(serverUrl)
const pageUrl = `${location.protocol}//${location.host}`
console.log("SERVER CHECK", url, pageUrl)
if (url != pageUrl && location.protocol == "https:" && this.sslCertified.includes(url) == false) {
if (url != pageUrl && url.length != 0 && location.protocol == "https:" && this.sslCertified.includes(url) == false) {
if (openTab) {
const value = window.confirm("MMVC Server is different from this page's origin. Open tab to open ssl connection. OK? (You can close the opened tab after ssl connection succeed.)");
if (value) {

4
server/misc/log_control.py
View File

@ -9,7 +9,7 @@ import logging
# logger = logging.getLogger("uvicorn.error")
# logger.addFilter(UvicornSuppressFilter())
# # logger.propagate = False
# logger = logging.getLogger("multipart.multipart")
# logger.propagate = False
logger = logging.getLogger("multipart.multipart")
logger.propagate = False
logging.getLogger('asyncio').setLevel(logging.WARNING)

23
server/restapi/MMVC_Rest_VoiceChanger.py
View File

@ -1,5 +1,6 @@
import base64, struct
import numpy as np
import traceback
from fastapi import APIRouter
from fastapi.encoders import jsonable_encoder
@ -10,14 +11,7 @@ from pydantic import BaseModel
import threading
class VoiceModel(BaseModel):
gpu: int
srcId: int
dstId: int
timestamp: int
convertChunkNum: int
crossFadeLowerValue: float
crossFadeOffsetRate:float
crossFadeEndRate:float
buffer: str
class MMVC_Rest_VoiceChanger:
@ -30,14 +24,7 @@ class MMVC_Rest_VoiceChanger:
def test(self, voice: VoiceModel):
try:
gpu = voice.gpu
srcId = voice.srcId
dstId = voice.dstId
timestamp = voice.timestamp
convertChunkNum = voice.convertChunkNum
crossFadeLowerValue = voice.crossFadeLowerValue
crossFadeOffsetRate = voice.crossFadeOffsetRate
crossFadeEndRate = voice.crossFadeEndRate
buffer = voice.buffer
wav = base64.b64decode(buffer)
@ -51,17 +38,12 @@ class MMVC_Rest_VoiceChanger:
# unpackedData.astype(np.int16))
self.tlock.acquire()
changedVoice = self.voiceChangerManager.changeVoice(
gpu, srcId, dstId, timestamp, convertChunkNum, crossFadeLowerValue, crossFadeOffsetRate, crossFadeEndRate, unpackedData)
changedVoice = self.voiceChangerManager.changeVoice( unpackedData)
self.tlock.release()
changedVoiceBase64 = base64.b64encode(changedVoice).decode('utf-8')
data = {
"gpu": gpu,
"srcId": srcId,
"dstId": dstId,
"timestamp": timestamp,
"convertChunkNum": voice.convertChunkNum,
"changedVoiceBase64": changedVoiceBase64
}
@ -71,6 +53,7 @@ class MMVC_Rest_VoiceChanger:
except Exception as e:
print("REQUEST PROCESSING!!!! EXCEPTION!!!", e)
print(traceback.format_exc())
self.tlock.release()
return str(e)

82
server/voice_changer/VoiceChanger.py
View File

@ -86,19 +86,25 @@ class VoiceChanger():
self.onnx_session = None
def destroy(self):
del self.net_g
del self.onnx_session
if hasattr(self, "net_g"):
del self.net_g
if hasattr(self, "onnx_session"):
del self.onnx_session
def get_info(self):
data = asdict(self.settings)
data["providers"] = self.onnx_session.get_providers() if hasattr(self, "onnx_session") else ""
data["providers"] = self.onnx_session.get_providers() if self.onnx_session != None else ""
files = ["config_file", "pyTorch_model_file", "onnx_model_file"]
for f in files:
data[f] = os.path.basename(data[f])
if os.path.exists(f):
data[f] = os.path.basename(data[f])
else:
data[f] = ""
return data
def update_setteings(self, key:str, val:any):
if key == "onnxExecutionProvider":
if key == "onnxExecutionProvider" and self.onnx_session != None:
if val == "CUDAExecutionProvider":
provider_options=[{'device_id': self.settings.gpu}]
self.onnx_session.set_providers(providers=[val], provider_options=provider_options)
@ -107,7 +113,7 @@ class VoiceChanger():
return self.get_info()
elif key in self.settings.intData:
setattr(self.settings, key, int(val))
if key == "gpu" and val >= 0 and val < self.gpu_num and hasattr(self, "onnx_session"):
if key == "gpu" and val >= 0 and val < self.gpu_num and self.onnx_session != None:
providers = self.onnx_session.get_providers()
print("Providers::::", providers)
if "CUDAExecutionProvider" in providers:
@ -176,35 +182,40 @@ class VoiceChanger():
def _onnx_inference(self, data, inputSize):
if hasattr(self, 'onnx_session'):
x, x_lengths, spec, spec_lengths, y, y_lengths, sid_src = [x for x in data]
sid_tgt1 = torch.LongTensor([self.settings.dstId])
# if spec.size()[2] >= 8:
audio1 = self.onnx_session.run(
["audio"],
{
"specs": spec.numpy(),
"lengths": spec_lengths.numpy(),
"sid_src": sid_src.numpy(),
"sid_tgt": sid_tgt1.numpy()
})[0][0,0] * self.hps.data.max_wav_value
if hasattr(self, 'np_prev_audio1') == True:
prev = self.np_prev_audio1[-1*inputSize:]
cur = audio1[-2*inputSize:-1*inputSize]
# print(prev.shape, self.np_prev_strength.shape, cur.shape, self.np_cur_strength.shape)
powered_prev = prev * self.np_prev_strength
powered_cur = cur * self.np_cur_strength
result = powered_prev + powered_cur
#result = prev * self.np_prev_strength + cur * self.np_cur_strength
else:
cur = audio1[-2*inputSize:-1*inputSize]
result = cur
self.np_prev_audio1 = audio1
return result
if hasattr(self, "onnx_session") == False or self.onnx_session == None:
print("[Voice Changer] No ONNX session.")
return np.zeros(1).astype(np.int16)
x, x_lengths, spec, spec_lengths, y, y_lengths, sid_src = [x for x in data]
sid_tgt1 = torch.LongTensor([self.settings.dstId])
# if spec.size()[2] >= 8:
audio1 = self.onnx_session.run(
["audio"],
{
"specs": spec.numpy(),
"lengths": spec_lengths.numpy(),
"sid_src": sid_src.numpy(),
"sid_tgt": sid_tgt1.numpy()
})[0][0,0] * self.hps.data.max_wav_value
if hasattr(self, 'np_prev_audio1') == True:
prev = self.np_prev_audio1[-1*inputSize:]
cur = audio1[-2*inputSize:-1*inputSize]
# print(prev.shape, self.np_prev_strength.shape, cur.shape, self.np_cur_strength.shape)
powered_prev = prev * self.np_prev_strength
powered_cur = cur * self.np_cur_strength
result = powered_prev + powered_cur
#result = prev * self.np_prev_strength + cur * self.np_cur_strength
else:
raise ValueError(ERROR_NO_ONNX_SESSION, "No ONNX Session.")
cur = audio1[-2*inputSize:-1*inputSize]
result = cur
self.np_prev_audio1 = audio1
return result
def _pyTorch_inference(self, data, inputSize):
if hasattr(self, "net_g") == False or self.net_g ==None:
print("[Voice Changer] No pyTorch session.")
return np.zeros(1).astype(np.int16)
if self.settings.gpu < 0 or self.gpu_num == 0:
with torch.no_grad():
x, x_lengths, spec, spec_lengths, y, y_lengths, sid_src = [x.cpu() for x in data]
@ -281,8 +292,11 @@ class VoiceChanger():
except Exception as e:
print("VC PROCESSING!!!! EXCEPTION!!!", e)
print(traceback.format_exc())
del self.np_prev_audio1
del self.prev_audio1
if hasattr(self, "np_prev_audio1"):
del self.np_prev_audio1
if hasattr(self, "prev_audio1"):
del self.prev_audio1
return np.zeros(1).astype(np.int16)
result = result.astype(np.int16)
# print("on_request result size:",result.shape)