Feature Request: Add SenseVoice for faster voice-to-notes

[LauraGPT]

Feature Request

notesGPT converts voice recordings to notes. SenseVoice from FunASR would make the transcription step faster and add unique capabilities.

Why

• 170x real-time on GPU — non-autoregressive, much faster than Whisper for transcription
• 50+ languages with automatic language detection
• Emotion detection — tag notes with speaker emotion (useful for meeting notes)
• Audio event detection — detect laughter, applause, music in the recording
• OpenAI-compatible API — drop-in replacement:

pip install funasr
funasr-server --device cuda
# Same /v1/audio/transcriptions endpoint as OpenAI

Self-hosted Option

For users who want fully local/private note-taking, FunASR runs entirely offline after model download — no API keys or cloud services needed.

Happy to help with integration!