filliptmandClaude Opus 4.5
c3dd3f6255
- Add NovaSR backend for 16kHz -> 48kHz audio upsampling (~52KB model) - Integrate standalone fl_novasr module (no external NovaSR dependency) - Auto-download model from HuggingFace YatharthS/NovaSR - Force CPU on Mac MPS to avoid conv1d channel limit errors - Add model caching and proper device handling Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>