feat: copy and refacotor the code from DDSP_SVC

client/demo/package.json

@@ -62,6 +62,7 @@
     },
     "dependencies": {
         "@alexanderolsen/libsamplerate-js": "^2.1.0",
+        "@babel/runtime": "^7.23.4",
         "@dannadori/voice-changer-client-js": "^1.0.177",
         "@dannadori/voice-changer-js": "^1.0.2",
         "@dannadori/worker-manager": "^1.0.20",

client/demo/src/components/demo/904-3_FileUploader.tsx

@@ -93,12 +93,6 @@ export const FileUploaderScreen = (props: FileUploaderScreenProps) => {
                 return enough;
             } else if (setting.voiceChangerType == "DDSP-SVC") {
                 const enough =
-                    !!setting.files.find((x) => {
-                        return x.kind == "ddspSvcModel";
-                    }) &&
-                    !!setting.files.find((x) => {
-                        return x.kind == "ddspSvcModelConfig";
-                    }) &&
                     !!setting.files.find((x) => {
                         return x.kind == "ddspSvcDiffusion";
                     }) &&

client/demo/src/components/demo/components2/100_ModelSlotArea.tsx

@@ -4,6 +4,7 @@ import { useGuiState } from "../001_GuiStateProvider";
 import { useMessageBuilder } from "../../../hooks/useMessageBuilder";
 import { FontAwesomeIcon } from "@fortawesome/react-fontawesome";
 import { useAppRoot } from "../../../001_provider/001_AppRootProvider";
+import { DDSPSVCModelSlot } from "@dannadori/voice-changer-client-js";
 
 export type ModelSlotAreaProps = {};
 
@@ -32,8 +33,8 @@ export const ModelSlotArea = (_props: ModelSlotAreaProps) => {
             sortType == "slot"
                 ? serverSetting.serverSetting.modelSlots
                 : serverSetting.serverSetting.modelSlots.slice().sort((a, b) => {
-                      return a.name.localeCompare(b.name);
-                  });
+                    return a.name.localeCompare(b.name);
+                });
 
         return modelSlots
             .map((x, index) => {

server/requirements.txt

@@ -27,3 +27,6 @@ websockets==11.0.2
 sounddevice==0.4.6
 dataclasses_json==0.5.7
 onnxsim==0.4.28
+diffusion==6.10.1
+transformers==4.35.2
+# praat-parselmouth==0.4.3

server/voice_changer/DDSP_SVC/DDSP_SVC.py

@@ -63,7 +63,7 @@ def __init__(self, params: VoiceChangerParams, slotInfo: DDSPSVCModelSlot):
         self.params = params
         self.settings = DDSP_SVCSettings()
         self.svc_model: SvcDDSP = SvcDDSP()
-        self.diff_model: DiffGtMel = DiffGtMel()
+        # self.diff_model: DiffGtMel = DiffGtMel()
 
         self.svc_model.setVCParams(params)
         EmbedderManager.initialize(params)
@@ -81,19 +81,19 @@ def initialize(self):
             str(self.slotInfo.slotIndex),
             "model",
             self.slotInfo.modelFile,
-        )
+        ) if self.slotInfo.modelFile != 'builtin' else 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)
-        self.svc_model.update_model(modelPath, self.device)
-        self.diff_model = DiffGtMel(device=self.device)
-        self.diff_model.flush_model(diffPath, ddsp_config=self.svc_model.args)
+        self.svc_model.update_model(modelPath, diffPath , self.device)
+        # self.diff_model = DiffGtMel(device=self.device)
+        # self.diff_model.flush_model(diffPath, ddsp_config=self.svc_model.args)
 
     def update_settings(self, key: str, val: int | float | str):
         if key in self.settings.intData:
@@ -161,19 +161,19 @@ def _pyTorch_inference(self, data):
             pitch_adjust=self.settings.tran,
             use_spk_mix=False,
             spk_mix_dict=None,
-            use_enhancer=True if self.settings.useEnhancer == 1 else False,
+            # use_enhancer=True if self.settings.useEnhancer == 1 else False,
             pitch_extractor_type=self.settings.f0Detector,
             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,
-            diff_model=self.diff_model,
+            # 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_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,
+            diff_method=self.settings.diffMethod,
             k_step=self.settings.kStep,  # TBD なにこれ？
             diff_silence=True
             if self.settings.useDiffSilence == 1

server/voice_changer/DDSP_SVC/DDSP_SVCModelSlotGenerator.py

@@ -17,6 +17,10 @@ def loadModel(cls, props: LoadModelParams):
                 slotInfo.diffConfigFile = file.name
             elif file.kind == "ddspSvcDiffusion":
                 slotInfo.diffModelFile = file.name
+        if slotInfo.configFile == '':
+            slotInfo.configFile = 'builtin'
+        if slotInfo.modelFile == '':
+            slotInfo.modelFile = 'builtin'
         slotInfo.isONNX = slotInfo.modelFile.endswith(".onnx")
-        slotInfo.name = os.path.splitext(os.path.basename(slotInfo.modelFile))[0]
+        slotInfo.name = os.path.splitext(os.path.basename(slotInfo.diffModelFile))[0]
         return slotInfo

server/voice_changer/DDSP_SVC/DDSP_SVCSetting.py

@@ -11,13 +11,13 @@ class DDSP_SVCSettings:
     silentThreshold: float = 0.00001
     extraConvertSize: int = 1024 * 32
 
-    useEnhancer: int = 0
-    useDiff: int = 1
+    # useEnhancer: int = 0
+    # useDiff: int = 1
     # useDiffDpm: int = 0
     diffMethod: str = "dpm-solver"  # "pndm", "dpm-solver"
     useDiffSilence: int = 0
     diffAcc: int = 20
-    diffSpkId: int = 1
+    # diffSpkId: int = 1
     kStep: int = 120
     threshold: int = -45
 
@@ -28,12 +28,12 @@ class DDSP_SVCSettings:
         "dstId",
         "tran",
         "extraConvertSize",
-        "useEnhancer",
-        "useDiff",
+        # "useEnhancer",
+        # "useDiff",
         # "useDiffDpm",
         "useDiffSilence",
         "diffAcc",
-        "diffSpkId",
+        # "diffSpkId",
         "kStep",
     ]
     floatData = ["silentThreshold"]