Jamie Pine
|
2bc243f93e
|
Add TTS Provider Architecture plan
Solves GitHub 2GB limit + frequent update UX issues by splitting app into:
- Main app (~150MB): UI + backend logic + Whisper
- TTS Providers (plugins): Separate downloadable binaries
- pytorch-cpu (~300MB)
- pytorch-cuda (~2.4GB)
- mlx (~800MB, macOS)
- remote (connect to external server)
- openai (API wrapper)
Benefits:
- Main app under GitHub 2GB limit
- Updates don't require re-downloading providers
- User choice of compute backend
- External provider support for teams/cloud
- Future-proof extensibility
|
2026-01-31 02:09:42 -08:00 |
|