请问是否有针对戏曲的模型？感谢

[wtf2024]

Is there a model for Chinese opera? thank

[Derpiesaurus]

If it's just singing, the vocal stems of a lot of the MDX models do a pretty good job. I don't know about chinese specifically, but I don't see why it wouldn't work somewhat at least, since language aside, they are just human vocals.

[wtf2024]

If it's just singing, the vocal stems of a lot of the MDX models do a pretty good job. I don't know about chinese specifically, but I don't see why it wouldn't work somewhat at least, since language aside, they are just human vocals.

具体为什么我也不太清楚，是替家里长辈操作的。我尝试了数个模型，据观察在高频人声的时候，有的模型消除不了人声，有的会把声音降得很低，导致整体音量时高时低。而我用流行歌曲尝试的时候效果很好。并且我怀疑跟原文件的清晰度也一定关系，毕竟现在年轻人都不爱京剧了，很难找到高清的资源。 I don't know exactly why, it was done for the elders in the family. I tried several models, and it was observed that some models could not eliminate the human voice when the human voice was high-frequency, and some models would lower the voice very low, causing the overall volume to fluctuate from time to time. And it worked fine when I tried it with pop songs. And I suspect that it has something to do with the clarity of the original files, after all, young people don't like Peking opera anymore, so it's hard to find high-definition resources.

[TripleKiller666]

There is a cutoff of about 15hz at the top frequencies, so anything more in the middle is much more likely to be retained compared to very high pitched Voices.