EmeryLin

Could you share the trained models?

suggest a new v2 version code for faster sampling

4

v0 and v1 are slow when sample too much key-points, for example 4800, it cost more than hours, thus I code a new version for faster sampling. the code is...

a faster sampling version

语音分割的部分是否能考虑其他的方法？感觉目前语音分割的并不是很准确

1

例如使用基于python的AI模型来进行人声分割?或者利用http://www.zhiqubz.com/ 中的开源https://github.com/deezer/spleeter 人声提取技术将人声分割后再进行分割?