长音频声音克隆

[xipingL]

是否支持单个的长音频克隆，因为我看项目提供的是几秒钟的零克隆，如果提供的是单个3-5分钟带参考文本的长音频，那要怎么做到克隆效果，或者有什么推荐的方案或者项目吗？感谢！

[aluminumbox]

切句挑一个质量高的片段，或者自己提取一下embedding做一下平均

[xipingL]

切句挑一个质量高的片段，或者自己提取一下embedding做一下平均

提取embedding没问题，就是speech_token的提取会报错，还是说这个就不提取了

  File "C:\Users\root\Desktop\workspace\_tts\tts-collect\webui.py", line 89, in cosyvoice_gen
    for i, j in enumerate(cosyvoice.inference_zero_shot(text, prompt_text, prompt_speech_16k, stream=False)):
  File "C:\Users\root\Desktop\workspace\_tts\tts-collect\CosyVoice\cosyvoice\cli\cosyvoice.py", line 69, in inference_zero_shot
    model_input = self.frontend.frontend_zero_shot(i, prompt_text, prompt_speech_16k)
  File "C:\Users\root\Desktop\workspace\_tts\tts-collect\CosyVoice\cosyvoice\cli\frontend.py", line 150, in frontend_zero_shot
    speech_token, speech_token_len = self._extract_speech_token(prompt_speech_16k)
  File "C:\Users\root\Desktop\workspace\_tts\tts-collect\CosyVoice\cosyvoice\cli\frontend.py", line 83, in _extract_speech_token
    speech_token = self.speech_tokenizer_session.run(None,
  File "C:\ProgramData\miniconda3\envs\tts-show\lib\site-packages\onnxruntime\capi\onnxruntime_inference_collection.py", line 220, in run
    return self._sess.run(output_names, input_feed, run_options)
onnxruntime.capi.onnxruntime_pybind11_state.Fail: [ONNXRuntimeError] : 1 : FAIL : Non-zero status code returned while running Add node. Name:'/Add_2' Status Message: /Add_2: right operand cannot broadcast on dim 0 LeftShape: {1,2341,1280}, RightShape: {1500,1280}