speakerの自動検知

hpx502766238 · Jul 3, 2023 · d0224a5 · d0224a5
1 parent 70956fa
commit d0224a5
Show file tree

Hide file tree

Showing 2 changed files with 5 additions and 1 deletion.
diff --git a/server/voice_changer/RVC/RVC.py b/server/voice_changer/RVC/RVC.py
@@ -144,7 +144,7 @@ def inference(self, data):
 
         audio = torchaudio.functional.resample(audio, self.slotInfo.samplingRate, 16000, rolloff=0.99)
         repeat = 1 if self.settings.rvcQuality else 0
-        sid = 0
+        sid = self.settings.dstId
         f0_up_key = self.settings.tran
         index_rate = self.settings.indexRatio
         protect = self.settings.protect

diff --git a/server/voice_changer/RVC/RVCModelSlotGenerator.py b/server/voice_changer/RVC/RVCModelSlotGenerator.py
@@ -100,6 +100,10 @@ def _setInfoByPytorch(cls, slot: ModelSlot):
             if slot.embedder.endswith("768"):
                 slot.embedder = slot.embedder[:-3]
 
+            if "speaker_info" in cpt.keys():
+                for k, v in cpt["speaker_info"].items():
+                    slot.speakers[int(k)] = str(v)
+
             # if slot.embedder == EnumEmbedderTypes.hubert.value:
             #     slot.embedder = EnumEmbedderTypes.hubert
             # elif slot.embedder == EnumEmbedderTypes.contentvec.value: