FunCodec is a research-oriented toolkit for audio quantization and downstream applications, such as text-to-speech synthesis, music generation et.al.
370
stars
30
forks
source link
audio_codec-encodec-zh_en-general-16k-nq32ds640-pytorch模型，Huggingface和Modelscope下载的参数不同 #52
Open
yellowcomets opened 3 months ago
Huggingface下载的model.pth md5sum是4285b02283940535f4636787dc801dda，Modelscope下载的model.pth md5sum是277ad4e9033a825d78622d691df47b82，对同一音频进行编码得到的code也不相同，不是同一模型吗？