Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

可以增加一个训练中文TTS模型的效果好速度快的教程吗? #1709

Open
keepWhere opened this issue Jan 14, 2025 · 6 comments

Comments

@keepWhere
Copy link

按照教程使用baker_zh下的prepare.sh,运行的时候网络报错导致无法继续往下进行,而且MeloTTS相对效果还可以,但是使用了一个中文底模去训练的时候,报错了调用JNI时,Java_com_k2fsa_sherpa_onnx_OfflineTts_generateWithCallbackImpl这里出错了,但是看不到更多信息,但是在电脑端推理是正常的,请问可以提供更多的中文TTS训练的文档吗

@csukuangfj
Copy link
Collaborator

你要贴完整的log

@keepWhere
Copy link
Author

--------- beginning of system
--------- beginning of crash
2025-01-14 11:49:24.269 4605-4605 DEBUG pid-4605 A #1 pc 00000000001aa330 /data/app/~~DkIc6Llp2lWhEPs-X2Y_Ng==/com.k2fsa.sherpa.onnx-AOaxed59q29Sd5WRsfxxPg==/lib/arm64/libsherpa-onnx-jni.so (BuildId: 8765a0a8925b9626e874c1d7f7536e8da035f679)
2025-01-14 11:49:24.269 4605-4605 DEBUG pid-4605 A #2 pc 00000000001a9c28 /data/app/~~DkIc6Llp2lWhEPs-X2Y_Ng==/com.k2fsa.sherpa.onnx-AOaxed59q29Sd5WRsfxxPg==/lib/arm64/libsherpa-onnx-jni.so (BuildId: 8765a0a8925b9626e874c1d7f7536e8da035f679)
2025-01-14 11:49:24.269 4605-4605 DEBUG pid-4605 A #3 pc 00000000001a9ae8 /data/app/~~DkIc6Llp2lWhEPs-X2Y_Ng==/com.k2fsa.sherpa.onnx-AOaxed59q29Sd5WRsfxxPg==/lib/arm64/libsherpa-onnx-jni.so (BuildId: 8765a0a8925b9626e874c1d7f7536e8da035f679)
2025-01-14 11:49:24.269 4605-4605 DEBUG pid-4605 A #4 pc 00000000001c49c8 /data/app/~~DkIc6Llp2lWhEPs-X2Y_Ng==/com.k2fsa.sherpa.onnx-AOaxed59q29Sd5WRsfxxPg==/lib/arm64/libsherpa-onnx-jni.so (BuildId: 8765a0a8925b9626e874c1d7f7536e8da035f679)
2025-01-14 11:49:24.269 4605-4605 DEBUG pid-4605 A #5 pc 00000000001c493c /data/app/~~DkIc6Llp2lWhEPs-X2Y_Ng==/com.k2fsa.sherpa.onnx-AOaxed59q29Sd5WRsfxxPg==/lib/arm64/libsherpa-onnx-jni.so (__cxa_throw+128) (BuildId: 8765a0a8925b9626e874c1d7f7536e8da035f679)
2025-01-14 11:49:24.269 4605-4605 DEBUG pid-4605 A #6 pc 00000000001f0f98 /data/app/~~DkIc6Llp2lWhEPs-X2Y_Ng==/com.k2fsa.sherpa.onnx-AOaxed59q29Sd5WRsfxxPg==/lib/arm64/libsherpa-onnx-jni.so (BuildId: 8765a0a8925b9626e874c1d7f7536e8da035f679)
2025-01-14 11:49:24.269 4605-4605 DEBUG pid-4605 A #7 pc 00000000002ea254 /data/app/~~DkIc6Llp2lWhEPs-X2Y_Ng==/com.k2fsa.sherpa.onnx-AOaxed59q29Sd5WRsfxxPg==/lib/arm64/libsherpa-onnx-jni.so (BuildId: 8765a0a8925b9626e874c1d7f7536e8da035f679)
2025-01-14 11:49:24.269 4605-4605 DEBUG pid-4605 A #8 pc 00000000002c4afc /data/app/~~DkIc6Llp2lWhEPs-X2Y_Ng==/com.k2fsa.sherpa.onnx-AOaxed59q29Sd5WRsfxxPg==/lib/arm64/libsherpa-onnx-jni.so (BuildId: 8765a0a8925b9626e874c1d7f7536e8da035f679)
2025-01-14 11:49:24.269 4605-4605 DEBUG pid-4605 A #9 pc 00000000001d7aec /data/app/~~DkIc6Llp2lWhEPs-X2Y_Ng==/com.k2fsa.sherpa.onnx-AOaxed59q29Sd5WRsfxxPg==/lib/arm64/libsherpa-onnx-jni.so (Java_com_k2fsa_sherpa_onnx_OfflineTts_generateWithCallbackImpl+552) (BuildId: 8765a0a8925b9626e874c1d7f7536e8da035f679)
--------- beginning of main

image

上面的就是完整的logs了,也看不到更多的信息了,就是我自己用一个Bert-VITS2的中文底模,通过MeloTTS训练后,在android的sherpa-onnx上运行得到的日志结果

@csukuangfj
Copy link
Collaborator

log 还是不全。你可以用我们的c++ api测试

@keepWhere
Copy link
Author

image

我的是中文底模的MeloTTS,用上面的哪个来测试?

@keepWhere
Copy link
Author

或者,我想问的是有没有可以直接微调的中文TTS,微调完,可以直接导出onnx直接在sherpa-onnx上面推理的支持呢?

@csukuangfj
Copy link
Collaborator

我想问的是有没有可以直接微调的中文TTS

请看 icefall 的 baker_zh https://github.com/k2-fsa/icefall/tree/master/egs/baker_zh/TTS

它训练出来的模型,可以直接在 sherpa-onnx 里面用。例子为
https://k2-fsa.github.io/sherpa/onnx/tts/pretrained_models/matcha.html


我的是中文底模的MeloTTS,用上面的哪个来测试?

请看文档里,我们是如何测试 melo tts 模型的。

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants