add minimum convert size

client/demo/src/101_server_setting.tsx

@@ -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"

client/demo/src/104_convert_setting.tsx

@@ -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 = {

client/demo/src/105_advanced_setting.tsx

@@ -108,6 +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">
@@ -240,6 +254,7 @@ export const useAdvancedSetting = (props: UseAdvancedSettingProps): AdvancedSett
                 <div className="body-row divider"></div>
 
                 {convertChunkNumRow}
+                {minConvertSizeRow}
                 {crossFadeOverlapRateRow}
                 {crossFadeOffsetRateRow}
                 {crossFadeEndRateRow}
@@ -251,7 +266,7 @@ export const useAdvancedSetting = (props: UseAdvancedSettingProps): AdvancedSett
                 <div className="body-row divider"></div>
             </>
         )
-    }, [showAdvancedSetting, mmvcServerUrlRow, protocolRow, 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(() => {

client/demo/src/hooks/useServerSetting.ts

@@ -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
     /////////////
@@ -218,6 +229,7 @@ export const useServerSetting = (props: UseServerSettingProps): ServerSettingSta
         setSrcId,
         setDstId,
         setConvertChunkNum,
+        setMinConvertSize,
         setGpu,
         setCrossFadeOffsetRate,
         setCrossFadeEndRate,

client/lib/src/const.ts

@@ -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,

server/voice_changer/VoiceChanger.py

@@ -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)