search

willwade / tts-wrapper

TTS-Wrapper makes it easier to use text-to-speech APIs by providing a unified and easy-to-use interface.
MIT License
9 stars 2 forks source link

Update Sherpa onnx to use other models #31

Open willwade opened 4 weeks ago

willwade commented 4 weeks ago

DO you think we should create a dev function. It would a) get huggingface JSON b) generate JSON using createindex.py - c) merge both JSON and save to the repo. Remove any other json in this repo

sherpa-onnx then would getVoices with no token or model path - with ALL data in the local JSON. Actually speeding things up as the end user doesnt download this file.. rename local download path to not ~/mms but ~/sherpaonnx or something more relevant I think a user should be able to provide their own onnx path - and then settings for this generation too..

I have a feeling it may not work great. In part because my createindex.py was broken. It was calculating the language codes correctly of each model etc.

@willwade I added your comment in the PR so it would not get lost.

Chafid commented 3 days ago

relevant PR is https://github.com/willwade/tts-wrapper/pull/36