关于语音复刻 #118
Replies: 5 comments 8 replies
-
两个测试用的 spk 文件
|
Beta Was this translation helpful? Give feedback.
-
请问build 之后保存下来的 json 文件是传到tts 的音色(上传)那里吗,我上传之后显示load failed,报错如下 During handling of the above exception, another exception occurred: Traceback (most recent call last): |
Beta Was this translation helpful? Give feedback.
-
目前启动webui默认是chattts的模型,有启动cosyvoice和fishspeech模型webui的设置么,还是在施工中? |
Beta Was this translation helpful? Give feedback.
-
@zhzLuke96 fishspeech在api使用mona.spkv1.json,声音一阵男一阵女,音色也不对,是还不支持reference audio么 |
Beta Was this translation helpful? Give feedback.
-
请问是上传了音频和reference text 之后就可以直接使用吗?我使用楼上提供的json可以正常生成音频,但是我通过web不能正常提取音频(虽然返回了json,但是使用它生成的音频只有一秒杂音) |
Beta Was this translation helpful? Give feedback.
-
UPDATE 241111:
现目前所有模型都支持语音复刻
目前,用参考音频推理基本已经写完了,ChatTTS和CosyVoice已支持使用参考音频(reference)作为推理prompt
简单测试结果:
下面是测试的生成效果:
参考音频:
mona_in.mp4
合成结果:
mona_out1.mp4
由于 spk 文件不太好操作,所以重写了一个专门用于构建带有 sample audio/reference audio 说话人的页面(webui中)
ref issues #113 #111
Beta Was this translation helpful? Give feedback.
All reactions