What's Changed
- Document CPU-only installation (pip pulls ~3 GB of CUDA wheels by default) by @jflaflamme in #219
- Speed table: measured H100 rows off local storage by @manu-lm in #250
- lsd_depth_distill: 64 rows in one pass, drop the unused flow multiplier by @manu-lm in #253
- Finetune a released model into a new language by @manu-lm in #254
- Import Annotated from typing by @manu-lm in #256
- Run the training tests in CI, and make zip lengths explicit by @manu-lm in #257
- Run the training tests in CI by @manu-lm in #258
- Fix decimal splitting in sentence chunker by @preciousimo in #217
- Split tts_model into text chunking and model-state I/O by @manu-lm in #259
- Immutable defaults in seanet; document the two open() call sites by @manu-lm in #260
- CONTRIBUTING: encourage sharing trained models by @manu-lm in #266
- Add a --default-voice option to the serve command by @gabrieldemarmiesse in #271
- README: explain depth+CFG distillation by @manu-lm in #272
- Move text conditioning into modules/text_conditioner.py by @manu-lm in #275
- Use tanh GELU approximation in the backbone FFN by @manu-lm in #278
- Replace beartype with ty for type checking by @gabrieldemarmiesse-ai-agent in #283
- Add Hindi community model and configurable terminal punctuation by @Sedherthe in #288
- Add "pocket-tts-timestamped" project to README by @dpm63 in #290
- Add Korean 300M community model to README by @seastar105 in #291
- README: Czech example with a Czech sentence by @manu-lm in #292
New Contributors
- @jflaflamme made their first contribution in #219
- @preciousimo made their first contribution in #217
- @gabrieldemarmiesse-ai-agent made their first contribution in #263
- @Sedherthe made their first contribution in #288
- @dpm63 made their first contribution in #290
- @seastar105 made their first contribution in #291
Full Changelog: v3.0.2...v3.1.0