Apppublic

thotran9799/audio_samples

sourceHugging Faceupdated 27d agoView on Hugging Face

0likes

App README

Static comparison of pre-generated TTS clips: same sentence, several systems side by side. No model runs in this Space.

path	role
`index.html`, `app.js`, `style.css`	the UI
`catalog.json`	generated index of utterances and clips
`audio/<system>/<lang>/<voice>/`	generated audio copies
`build_space.py`	rebuilds `catalog.json` + `audio/` from `../audio_out`

bash

python build_space.py                 # re-read ../audio_out
python -m http.server 7871            # port 7860 is taken by the Voxtral gradio demo

Open http://127.0.0.1:7871 — opening index.html as file:// fails because catalog.json is fetched over HTTP.

Useful flags:

bash

python build_space.py --limit 3
python build_space.py --systems voxtral-4b-tts qwen3-tts-custom_voice
python build_space.py --dry-run

bash

git add -A && git commit -m "update samples" && git push

WAV files are committed directly; keep individual clips under 10 MB or add an LFS rule for *.wav in .gitattributes (needs git-lfs installed).