add sampling rate to rvc

client/demo/dist/assets/gui_settings/RVC.json

@@ -51,6 +51,10 @@
                 "options": {
                     "showFramework": true
                 }
+            },
+            {
+                "name": "modelSamplingRate",
+                "options": {}
             }
         ],
         "deviceSetting": [
@@ -160,6 +164,10 @@
             {
                 "name": "trancateNumThreshold",
                 "options": {}
+            },
+            {
+                "name": "rvcQuality",
+                "options": {}
             }
         ]
     },

client/demo/public/assets/gui_settings/RVC.json

@@ -51,6 +51,10 @@
                 "options": {
                     "showFramework": true
                 }
+            },
+            {
+                "name": "modelSamplingRate",
+                "options": {}
             }
         ],
         "deviceSetting": [
@@ -160,6 +164,10 @@
             {
                 "name": "trancateNumThreshold",
                 "options": {}
+            },
+            {
+                "name": "rvcQuality",
+                "options": {}
             }
         ]
     },

client/demo/src/components/demo/002_ComponentGenerator.tsx

@@ -34,6 +34,8 @@ import { CrossFadeEndRateRow, CrossFadeEndRateRowProps } from "./components/807_
 import { DownSamplingModeRow, DownSamplingModeRowProps } from "./components/808_DownSamplingModeRow"
 import { TrancateNumTresholdRow, TrancateNumTresholdRowProps } from "./components/809_TrancateNumTresholdRow"
 import { IndexRatioRow, IndexRatioRowProps } from "./components/609_IndexRatioRow"
+import { RVCQualityRow, RVCQualityRowProps } from "./components/810_RVCQuality"
+import { ModelSamplingRateRow, ModelSamplingRateRowProps } from "./components/303_ModelSamplingRateRow"
 
 export const catalog: { [key: string]: (props: any) => JSX.Element } = {}
 
@@ -61,6 +63,7 @@ const initialize = () => {
 
     addToCatalog("modelUploader", (props: ModelUploaderRowProps) => { return <ModelUploaderRow {...props} /> })
     addToCatalog("framework", (props: FrameworkRowProps) => { return <FrameworkRow {...props} /> })
+    addToCatalog("modelSamplingRate", (props: ModelSamplingRateRowProps) => { return <ModelSamplingRateRow {...props} /> })
 
 
     addToCatalog("audioInput", (props: AudioInputRowProps) => { return <AudioInputRow {...props} /> })
@@ -96,6 +99,8 @@ const initialize = () => {
     addToCatalog("crossFadeEndRate", (props: CrossFadeEndRateRowProps) => { return <CrossFadeEndRateRow {...props} /> })
     addToCatalog("downSamplingMode", (props: DownSamplingModeRowProps) => { return <DownSamplingModeRow {...props} /> })
     addToCatalog("trancateNumThreshold", (props: TrancateNumTresholdRowProps) => { return <TrancateNumTresholdRow {...props} /> })
+    addToCatalog("rvcQuality", (props: RVCQualityRowProps) => { return <RVCQualityRow {...props} /> })
+
 
 
 }

client/demo/src/components/demo/components/303_ModelSamplingRateRow.tsx

@@ -0,0 +1,43 @@
+import React, { useMemo } from "react"
+import { fileSelector, ModelSamplingRate } from "@dannadori/voice-changer-client-js"
+import { useAppState } from "../../../001_provider/001_AppStateProvider"
+
+export type ModelSamplingRateRowProps = {
+}
+
+export const ModelSamplingRateRow = (_props: ModelSamplingRateRowProps) => {
+    const appState = useAppState()
+
+    const modelSamplingRateRow = useMemo(() => {
+        const onModelSamplingRateChanged = (val: ModelSamplingRate) => {
+            appState.serverSetting.updateServerSettings({
+                ...appState.serverSetting.serverSetting,
+                modelSamplingRate: val
+            })
+        }
+
+        return (
+            <div className="body-row split-3-3-4 left-padding-1 guided">
+                <div className="body-item-title left-padding-2">Model Sampling Rate</div>
+                <div className="body-item-text">
+                    <div></div>
+                </div>
+                <div className="body-button-container">
+                    <select className="body-select" value={appState.serverSetting.serverSetting.modelSamplingRate} onChange={(e) => {
+                        onModelSamplingRateChanged(e.target.value as unknown as ModelSamplingRate)
+                    }}>
+                        {
+                            Object.values(ModelSamplingRate).map(x => {
+                                return <option key={x} value={x}>{x}</option>
+                            })
+                        }
+                    </select>
+
+
+                </div>
+            </div>
+        )
+    }, [appState.serverSetting.serverSetting, appState.serverSetting.updateServerSettings])
+
+    return modelSamplingRateRow
+}

client/demo/src/components/demo/components/810_RVCQuality.tsx

@@ -0,0 +1,31 @@
+import React, { useMemo } from "react"
+import { useAppState } from "../../../001_provider/001_AppStateProvider"
+
+export type RVCQualityRowProps = {
+}
+
+export const RVCQualityRow = (_props: RVCQualityRowProps) => {
+    const appState = useAppState()
+
+    const trancateNumTresholdRow = useMemo(() => {
+        const onRVCQualityChanged = (val: number) => {
+            appState.serverSetting.updateServerSettings({
+                ...appState.serverSetting.serverSetting,
+                rvcQuality: val
+            })
+        }
+        return (
+            <div className="body-row split-3-7 left-padding-1 guided">
+                <div className="body-item-title left-padding-1">RCV Quality</div>
+                <div className="body-input-container">
+                    <select value={appState.serverSetting.serverSetting.rvcQuality} onChange={(e) => { onRVCQualityChanged(Number(e.target.value)) }}>
+                        <option value="0" >low</option>
+                        <option value="1" >high</option>
+                    </select>
+                </div>
+            </div>
+        )
+    }, [appState.serverSetting.serverSetting, appState.serverSetting.updateServerSettings])
+
+    return trancateNumTresholdRow
+}

client/lib/src/const.ts

@@ -25,6 +25,14 @@ export const InputSampleRate = {
 } as const
 export type InputSampleRate = typeof InputSampleRate[keyof typeof InputSampleRate]
 
+export const ModelSamplingRate = {
+    "48000": 48000,
+    "40000": 40000,
+    "32000": 32000
+} as const
+export type ModelSamplingRate = typeof InputSampleRate[keyof typeof InputSampleRate]
+
+
 export const CrossFadeOverlapSize = {
     "1024": 1024,
     "2048": 2048,
@@ -79,6 +87,9 @@ export const ServerSettingKey = {
     "clusterInferRatio": "clusterInferRatio",
 
     "indexRatio": "indexRatio",
+    "rvcQuality": "rvcQuality",
+    "modelSamplingRate": "modelSamplingRate",
+
 
     "inputSampleRate": "inputSampleRate",
 } as const
@@ -109,6 +120,8 @@ export type VoiceChangerServerSetting = {
     clusterInferRatio: number // so-vits-svc
 
     indexRatio: number // RVC
+    rvcQuality: number // 0:low, 1:high
+    modelSamplingRate: ModelSamplingRate // 32000,40000,48000
 
     inputSampleRate: InputSampleRate
 }
@@ -147,6 +160,8 @@ export const DefaultServerSetting_MMVCv15: ServerInfo = {
     clusterInferRatio: 0,
 
     indexRatio: 0,
+    rvcQuality: 0,
+    modelSamplingRate: 48000,
 
     inputSampleRate: 24000,
 
@@ -181,6 +196,9 @@ export const DefaultServerSetting_MMVCv13: ServerInfo = {
     clusterInferRatio: 0,
 
     indexRatio: 0,
+    rvcQuality: 0,
+    modelSamplingRate: 48000,
+
 
     inputSampleRate: 24000,
 
@@ -219,6 +237,9 @@ export const DefaultServerSetting_so_vits_svc_40: ServerInfo = {
     clusterInferRatio: 0.1,
 
     indexRatio: 0,
+    rvcQuality: 0,
+    modelSamplingRate: 48000,
+
 
     inputSampleRate: 24000,
 
@@ -257,6 +278,9 @@ export const DefaultServerSetting_so_vits_svc_40_c: ServerInfo = {
     clusterInferRatio: 0.1,
 
     indexRatio: 0,
+    rvcQuality: 0,
+    modelSamplingRate: 48000,
+
 
     inputSampleRate: 24000,
 
@@ -294,6 +318,9 @@ export const DefaultServerSetting_so_vits_svc_40v2: ServerInfo = {
     clusterInferRatio: 0.1,
 
     indexRatio: 0,
+    rvcQuality: 0,
+    modelSamplingRate: 48000,
+
 
     inputSampleRate: 24000,
 
@@ -333,6 +360,8 @@ export const DefaultServerSetting_RVC: ServerInfo = {
     clusterInferRatio: 0.1,
 
     indexRatio: 0,
+    rvcQuality: 0,
+    modelSamplingRate: 48000,
 
     inputSampleRate: 24000,
 

server/voice_changer/RVC/RVC.py

@@ -28,7 +28,7 @@ from const import HUBERT_ONNX_MODEL_PATH
 import pyworld as pw
 
 from vc_infer_pipeline import VC
-from infer_pack.models import SynthesizerTrnMs256NSFsid, SynthesizerTrnMs256NSFsid_nono
+from infer_pack.models import SynthesizerTrnMs256NSFsid
 from fairseq import checkpoint_utils
 providers = ['OpenVINOExecutionProvider', "CUDAExecutionProvider", "DmlExecutionProvider", "CPUExecutionProvider"]
 
@@ -52,14 +52,15 @@ class RVCSettings():
     configFile: str = ""
 
     indexRatio: float = 0
-    quality: int = 0
+    rvcQuality: int = 0
+    modelSamplingRate: int = 48000
 
     speakers: dict[str, int] = field(
         default_factory=lambda: {}
     )
 
     # ↓mutableな物だけ列挙
-    intData = ["gpu", "dstId", "tran", "predictF0", "extraConvertSize", "quality"]
+    intData = ["gpu", "dstId", "tran", "predictF0", "extraConvertSize", "rvcQuality", "modelSamplingRate"]
     floatData = ["noiceScale", "silentThreshold", "indexRatio"]
     strData = ["framework", "f0Detector"]
 
@@ -82,7 +83,6 @@ class RVC:
         self.index_file = index_file
         self.is_half = is_half
 
-        self.tgt_sr = 40000
         try:
             hubert_path = self.params["hubert"]
             models, saved_cfg, task = checkpoint_utils.load_model_ensemble_and_task([hubert_path], suffix="",)
@@ -103,7 +103,7 @@ class RVC:
         # PyTorchモデル生成
         if pyTorch_model_file != None:
             cpt = torch.load(pyTorch_model_file, map_location="cpu")
-            self.tgt_sr = cpt["config"][-1]
+            self.settings.tgt_sr = cpt["config"][-1]
             net_g = SynthesizerTrnMs256NSFsid(*cpt["config"], is_half=self.is_half)
             net_g.eval()
             net_g.load_state_dict(cpt["weight"], strict=False)
@@ -113,14 +113,7 @@ class RVC:
 
         # ONNXモデル生成
         if onnx_model_file != None:
-            # self.onnx_session = ModelWrapper(onnx_model_file, is_half=True)
             self.onnx_session = ModelWrapper(onnx_model_file, is_half=self.is_half)
-            # input_info = self.onnx_session.get_inputs()
-            # for i in input_info:
-            #     print("input", i)
-            # output_info = self.onnx_session.get_outputs()
-            # for i in output_info:
-            #     print("output", i)
         return self.get_info()
 
     def update_setteings(self, key: str, val: any):
@@ -170,7 +163,7 @@ class RVC:
         return data
 
     def get_processing_sampling_rate(self):
-        return self.tgt_sr
+        return self.settings.tgt_sr
         # return 24000
 
     def generate_input(self, newData: any, inputSize: int, crossfadeSize: int):
@@ -215,13 +208,13 @@ class RVC:
         convertSize = data[1]
         vol = data[2]
 
-        audio = resampy.resample(audio, self.tgt_sr, 16000)
+        audio = resampy.resample(audio, self.settings.tgt_sr, 16000)
 
         if vol < self.settings.silentThreshold:
             return np.zeros(convertSize).astype(np.int16)
 
         with torch.no_grad():
-            vc = VC(self.tgt_sr, dev, self.is_half)
+            vc = VC(self.settings.tgt_sr, dev, self.is_half)
             sid = 0
             times = [0, 0, 0]
             f0_up_key = self.settings.tran
@@ -255,14 +248,14 @@ class RVC:
         convertSize = data[1]
         vol = data[2]
         print("audio len 02,", len(audio))
-        audio = resampy.resample(audio, self.tgt_sr, 16000)
+        audio = resampy.resample(audio, self.settings.tgt_sr, 16000)
         print("audio len 03,", len(audio))
 
         if vol < self.settings.silentThreshold:
             return np.zeros(convertSize).astype(np.int16)
 
         with torch.no_grad():
-            vc = VC(self.tgt_sr, dev, self.is_half)
+            vc = VC(self.settings.tgt_sr, dev, self.is_half)
             sid = 0
             times = [0, 0, 0]
             f0_up_key = self.settings.tran