fastfading
fastfading
sparktts mac m4 上效果很好。 速度很快。 使用很方便， 不用像gptsovits 那样训练数据，或者找人家训练好的， 很麻烦。 另外多段语音之间的几乎没有间隔,两段语音之间的衔接过于紧凑. 这个能不能 优化一下
I have got https://github.com/remsky/Kokoro-FastAPI run locall , Question is: how to use that api in open llm vtuber
vs sensevoice and whisper ， which is the best ? https://zhuanlan.zhihu.com/p/1892898181790007705  looks good. can some one make it work on mac ?
实际测试结果 mac m1 small model mps auto detected mode 中文的效果勉强可用但是英文完全不行 the quality for chinese is good but for english is not as good as we expected . may related to...
https://github.com/culdo/web-mmd
related pull request https://github.com/Open-LLM-VTuber/Open-LLM-VTuber/pull/220/files
you need kokoro 1.1 zh
just use. https://github.com/remsky/Kokoro-FastAPI support cuda and mac mlx gpu question is how to use this api in open llm vtuber may be we need an common api interface
Interesting, could you provide a screenshot?