Problem with piper-tts models

[mah92]

Hi
I am trying to test SherpaOnnxTts on android.
Onnx lessac model from pretrained_models works but models from piper-voices repository like lessac not works (the program just hangs on startup). What conversion do you perform on piper models to work?
I use static libonnx(sherpa-onnx-static-link-onnxruntime-1.10.35.aar), can it be the problem source?

[mah92]

2. When converting a piper ckpt file, what steps are recommended?
3. Then how can I add a voice to sherpa repo? Should I add an issue or change somewhere and add a pull request?

[csukuangfj]

Have you ever read the sherpa-onnx doc for converting models from piper?

Also, when you encounter errors, please describe in detail how to reproduce them and post the complete logs. Otherwise, we cannot help you.

[mablue]

dont forget metadata generation
https://k2-fsa.github.io/sherpa/onnx/tts/piper.html

[mah92]

Thank you. It was the problem. I still have one question(which I did not find in the documents)
How can I add a voice to sherpa repo? Should I add an issue or change somewhere and add a pull request?

[csukuangfj]

Thank you. It was the problem. I still have one question(which I did not find in the documents)
How can I add a voice to sherpa repo? Should I add an issue or change somewhere and add a pull request?

Do you mean to add a new piper tts model?