feat: openvoice重构，添加语音克隆功能

utils/tts/openvoice_utils.py

@@ -260,6 +260,11 @@ class TextToSpeech:
             audio: tensor
             sr: 生成音频的采样速率
         """
+        if source_se is not None:
+            source_se = torch.tensor(source_se.astype(np.float32)).to(self.device)
+        if target_se is not None:
+            target_se = torch.tensor(target_se.astype(np.float32)).to(self.device)
+            
         if source_se is None:
             source_se = self.source_se
         if target_se is None:
@@ -319,7 +324,6 @@ class TextToSpeech:
             tts_sr = self.base_tts_model.hps.data.sampling_rate
             converter_sr = self.tone_color_converter.hps.data.sampling_rate
             audio = F.resample(audio, tts_sr, converter_sr)
-            print(audio.dtype)
             audio, sr = self._convert_tone(audio,
                                            source_se=source_se,
                                            target_se=target_se,