bugfix: beatrice load

w-okada · Nov 8, 2023 · d03132d · d03132d
1 parent 3512bbb
commit d03132d
Show file tree

Hide file tree

Showing 12 changed files with 525 additions and 310 deletions.
diff --git a/client/demo/build-voice-changer-js.sh b/client/demo/build-voice-changer-js.sh
@@ -4,7 +4,6 @@
 # cp -r ~/git-work/voice-changer-js/lib/package.json node_modules/@dannadori/voice-changer-js/
 # cp -r ~/git-work/voice-changer-js/lib/dist node_modules/@dannadori/voice-changer-js/
 
-
 cd ~/git-work/voice-changer-js/lib/ ; npm run build:prod; cd -
 rm -rf node_modules/@dannadori/voice-changer-js
 mkdir -p node_modules/@dannadori/voice-changer-js/dist

diff --git a/client/demo/src/001_provider/001_AppStateProvider.tsx b/client/demo/src/001_provider/001_AppStateProvider.tsx
@@ -1,12 +1,11 @@
 import { ClientState } from "@dannadori/voice-changer-client-js";
+import { VoiceChangerJSClient } from "@dannadori/voice-changer-js";
 import React, { useContext, useEffect, useRef } from "react";
 import { ReactNode } from "react";
 import { useVCClient } from "../001_globalHooks/001_useVCClient";
 import { useAppRoot } from "./001_AppRootProvider";
 import { useMessageBuilder } from "../hooks/useMessageBuilder";
 
-import { VoiceChangerJSClient } from "./VoiceChangerJSClient";
-
 type Props = {
     children: ReactNode;
 };
@@ -58,22 +57,34 @@ export const AppStateProvider = ({ children }: Props) => {
 
     // useEffect(() => {
     //     if (clientState.clientState.initialized) {
+    //         const baseUrl = "https://192.168.0.247:18888";
+    //         // const modelUrl = `${baseUrl}/models/rvc2v_40k_f0_24000.bin`;
+    //         // const modelUrl = `${baseUrl}/models/rvc2v_40k_nof0_24000.bin`;
+    //         // const modelUrl = `${baseUrl}/models/rvc2v_16k_f0_24000.bin`;
+    //         // const modelUrl = `${baseUrl}/models/rvcv2_amitaro_v2_40k_f0_24000.bin`;
+    //         // const modelUrl = `${baseUrl}/models/rvcv2_amitaro_v2_40k_nof0_24000.bin`;
+    //         // const modelUrl = `${baseUrl}/models/rvcv2_amitaro_v2_32k_f0_24000.bin`;
+    //         // const modelUrl = `${baseUrl}/models/rvcv2_amitaro_v2_32k_nof0_24000.bin`;
+
+    //         // const modelUrl = `${baseUrl}/models/rvcv1_amitaro_v1_32k_f0_24000.bin`;
+    //         const modelUrl = `${baseUrl}/models/rvcv1_amitaro_v1_32k_nof0_24000.bin`;
+    //         // const modelUrl = `${baseUrl}/models/rvcv1_amitaro_v1_40k_f0_24000.bin`;
+    //         // const modelUrl = `${baseUrl}/models/rvcv1_amitaro_v1_40k_nof0_24000.bin`;
+
     //         voiceChangerJSClient.current = new VoiceChangerJSClient();
-    //         voiceChangerJSClient.current.initialize();
+    //         voiceChangerJSClient.current.initialize(
+    //             {
+    //                 baseUrl: baseUrl,
+    //                 inputSamplingRate: 48000,
+    //                 outputSamplingRate: 48000,
+    //             },
+    //             modelUrl,
+    //         );
     //         clientState.clientState.setInternalAudioProcessCallback({
     //             processAudio: async (data: Uint8Array) => {
-    //                 console.log("[CLIENTJS] start --------------------------------------");
     //                 const audioF32 = new Float32Array(data.buffer);
     //                 const converted = await voiceChangerJSClient.current!.convert(audioF32);
-
-    //                 let audio_int16_out = new Int16Array(converted.length);
-    //                 for (let i = 0; i < converted.length; i++) {
-    //                     audio_int16_out[i] = converted[i] * 32768.0;
-    //                 }
-    //                 const res = new Uint8Array(audio_int16_out.buffer);
-    //                 console.log("AUDIO::::audio_int16_out", audio_int16_out);
-
-    //                 console.log("[CLIENTJS] end --------------------------------------");
+    //                 const res = new Uint8Array(converted.buffer);
     //                 return res;
     //             },
     //         });

diff --git a/client/demo/src/001_provider/VoiceChangerJSClient.ts b/client/demo/src/001_provider/VoiceChangerJSClient.ts
diff --git a/client/demo/webpack.common.js b/client/demo/webpack.common.js
@@ -55,20 +55,44 @@ module.exports = {
             patterns: [{ from: "public/favicon.ico", to: "favicon.ico" }],
         }),
 
+        new CopyPlugin({
+            patterns: [{ from: "./node_modules/@dannadori/voice-changer-js/dist/ort-wasm-simd.wasm", to: "ort-wasm-simd.wasm" }],
+        }),
         // new CopyPlugin({
-        //     patterns: [{ from: "./node_modules/@dannadori/voice-changer-js/dist/ort-wasm-simd.wasm", to: "ort-wasm-simd.wasm" }],
+        //     patterns: [{ from: "./node_modules/@dannadori/voice-changer-js/dist/tfjs-backend-wasm-simd.wasm", to: "tfjs-backend-wasm-simd.wasm" }],
         // }),
         // new CopyPlugin({
         //     patterns: [{ from: "./node_modules/@dannadori/voice-changer-js/dist/process.js", to: "process.js" }],
         // }),
         // new CopyPlugin({
-        //     patterns: [{ from: "public/models/emb_pit_24000.bin", to: "models/emb_pit_24000.bin" }],
+        //     patterns: [{ from: "public/models/rvcv2_emb_pit_24000.bin", to: "models/rvcv2_emb_pit_24000.bin" }],
+        // }),
+        // new CopyPlugin({
+        //     patterns: [{ from: "public/models/rvcv2_amitaro_v2_32k_f0_24000.bin", to: "models/rvcv2_amitaro_v2_32k_f0_24000.bin" }],
+        // }),
+        // new CopyPlugin({
+        //     patterns: [{ from: "public/models/rvcv2_amitaro_v2_32k_nof0_24000.bin", to: "models/rvcv2_amitaro_v2_32k_nof0_24000.bin" }],
+        // }),
+        // new CopyPlugin({
+        //     patterns: [{ from: "public/models/rvcv2_amitaro_v2_40k_f0_24000.bin", to: "models/rvcv2_amitaro_v2_40k_f0_24000.bin" }],
+        // }),
+        // new CopyPlugin({
+        //     patterns: [{ from: "public/models/rvcv2_amitaro_v2_40k_nof0_24000.bin", to: "models/rvcv2_amitaro_v2_40k_nof0_24000.bin" }],
+        // }),
+        // new CopyPlugin({
+        //     patterns: [{ from: "public/models/rvcv1_emb_pit_24000.bin", to: "models/rvcv1_emb_pit_24000.bin" }],
+        // }),
+        // new CopyPlugin({
+        //     patterns: [{ from: "public/models/rvcv1_amitaro_v1_32k_f0_24000.bin", to: "models/rvcv1_amitaro_v1_32k_f0_24000.bin" }],
+        // }),
+        // new CopyPlugin({
+        //     patterns: [{ from: "public/models/rvcv1_amitaro_v1_32k_nof0_24000.bin", to: "models/rvcv1_amitaro_v1_32k_nof0_24000.bin" }],
         // }),
         // new CopyPlugin({
-        //     patterns: [{ from: "public/models/rvc2v_24000.bin", to: "models/rvc2v_24000.bin" }],
+        //     patterns: [{ from: "public/models/rvcv1_amitaro_v1_40k_f0_24000.bin", to: "models/rvcv1_amitaro_v1_40k_f0_24000.bin" }],
         // }),
         // new CopyPlugin({
-        //     patterns: [{ from: "public/models/rvc2vnof0_24000.bin", to: "models/rvc2vnof0_24000.bin" }],
+        //     patterns: [{ from: "public/models/rvcv1_amitaro_v1_40k_nof0_24000.bin", to: "models/rvcv1_amitaro_v1_40k_nof0_24000.bin" }],
         // }),
     ],
 };
diff --git a/server/voice_changer/DDSP_SVC/DDSP_SVC.py b/server/voice_changer/DDSP_SVC/DDSP_SVC.py
@@ -20,7 +20,7 @@
 
 from .models.diffusion.infer_gt_mel import DiffGtMel
 
-from voice_changer.utils.VoiceChangerModel import AudioInOut
+from voice_changer.utils.VoiceChangerModel import AudioInOut, VoiceChangerModel
 from voice_changer.utils.VoiceChangerParams import VoiceChangerParams
 from voice_changer.DDSP_SVC.DDSP_SVCSetting import DDSP_SVCSettings
 from voice_changer.RVC.embedder.EmbedderManager import EmbedderManager
@@ -44,15 +44,20 @@ def phase_vocoder(a, b, fade_out, fade_in):
     deltaphase = deltaphase - 2 * np.pi * torch.floor(deltaphase / 2 / np.pi + 0.5)
     w = 2 * np.pi * torch.arange(n // 2 + 1).to(a) + deltaphase
     t = torch.arange(n).unsqueeze(-1).to(a) / n
-    result = a * (fade_out**2) + b * (fade_in**2) + torch.sum(absab * torch.cos(w * t + phia), -1) * fade_out * fade_in / n
+    result = (
+        a * (fade_out**2)
+        + b * (fade_in**2)
+        + torch.sum(absab * torch.cos(w * t + phia), -1) * fade_out * fade_in / n
+    )
     return result
 
 
-class DDSP_SVC:
+class DDSP_SVC(VoiceChangerModel):
     initialLoad: bool = True
 
     def __init__(self, params: VoiceChangerParams, slotInfo: DDSPSVCModelSlot):
         print("[Voice Changer] [DDSP-SVC] Creating instance ")
+        self.voiceChangerType = "DDSP-SVC"
         self.deviceManager = DeviceManager.get_instance()
         self.gpu_num = torch.cuda.device_count()
         self.params = params
@@ -71,8 +76,18 @@ def __init__(self, params: VoiceChangerParams, slotInfo: DDSPSVCModelSlot):
     def initialize(self):
         self.device = self.deviceManager.getDevice(self.settings.gpu)
         vcparams = VoiceChangerParamsManager.get_instance().params
-        modelPath = os.path.join(vcparams.model_dir, str(self.slotInfo.slotIndex), "model",  self.slotInfo.modelFile)
-        diffPath = os.path.join(vcparams.model_dir, str(self.slotInfo.slotIndex), "diff", self.slotInfo.diffModelFile)
+        modelPath = os.path.join(
+            vcparams.model_dir,
+            str(self.slotInfo.slotIndex),
+            "model",
+            self.slotInfo.modelFile,
+        )
+        diffPath = os.path.join(
+            vcparams.model_dir,
+            str(self.slotInfo.slotIndex),
+            "diff",
+            self.slotInfo.diffModelFile,
+        )
 
         self.svc_model = SvcDDSP()
         self.svc_model.setVCParams(self.params)
@@ -112,11 +127,15 @@ def generate_input(
         # newData = newData.astype(np.float32)
 
         if self.audio_buffer is not None:
-            self.audio_buffer = np.concatenate([self.audio_buffer, newData], 0)  # 過去のデータに連結
+            self.audio_buffer = np.concatenate(
+                [self.audio_buffer, newData], 0
+            )  # 過去のデータに連結
         else:
             self.audio_buffer = newData
 
-        convertSize = inputSize + crossfadeSize + solaSearchFrame + self.settings.extraConvertSize
+        convertSize = (
+            inputSize + crossfadeSize + solaSearchFrame + self.settings.extraConvertSize
+        )
 
         # if convertSize % self.hop_size != 0:  # モデルの出力のホップサイズで切り捨てが発生するので補う。
         #     convertSize = convertSize + (self.hop_size - (convertSize % self.hop_size))
@@ -147,15 +166,18 @@ def _pyTorch_inference(self, data):
             f0_min=50,
             f0_max=1100,
             # safe_prefix_pad_length=0,  # TBD なにこれ？
-            safe_prefix_pad_length=self.settings.extraConvertSize / self.svc_model.args.data.sampling_rate,
+            safe_prefix_pad_length=self.settings.extraConvertSize
+            / self.svc_model.args.data.sampling_rate,
             diff_model=self.diff_model,
             diff_acc=self.settings.diffAcc,  # TBD なにこれ？
             diff_spk_id=self.settings.diffSpkId,
             diff_use=True if self.settings.useDiff == 1 else False,
             # diff_use_dpm=True if self.settings.useDiffDpm == 1 else False,  # TBD なにこれ？
             method=self.settings.diffMethod,
             k_step=self.settings.kStep,  # TBD なにこれ？
-            diff_silence=True if self.settings.useDiffSilence == 1 else False,  # TBD なにこれ？
+            diff_silence=True
+            if self.settings.useDiffSilence == 1
+            else False,  # TBD なにこれ？
         )
 
         return _audio.cpu().numpy() * 32768.0
@@ -182,5 +204,4 @@ def __del__(self):
                 pass
 
     def get_model_current(self):
-        return [
-        ]
+        return []