Feature Request: Add FunASR / SenseVoice as STT backend

[LauraGPT]

Summary

It would be great to support FunASR and SenseVoice as speech-to-text backends in LiveKit Agents.

Why

• FunASR is an open-source industrial-grade speech recognition toolkit with real-time streaming support, punctuation restoration, and inverse text normalization built-in. It supports 50+ languages and can run fully offline.
• SenseVoice is a speech foundation model excelling in multilingual speech recognition, emotion recognition, and audio event detection — with ultra-low latency (~70ms for 10s audio via SenseVoice-Small).
• Both are Apache 2.0 licensed and available on HuggingFace.

Comparison with Whisper

Feature	FunASR/SenseVoice	Whisper
Real-time streaming	✅ Native	❌ Chunked
Punctuation & ITN	✅ Built-in	❌ Requires post-processing
Latency	~70ms (SenseVoice-Small)	Higher
Languages	50+	99
License	Apache 2.0	MIT

References

• FunASR GitHub: https://github.com/modelscope/FunASR (16K+ stars)
• SenseVoice GitHub: https://github.com/FunAudioLLM/SenseVoice (8K+ stars)
• Fun-ASR-Nano (HuggingFace native): https://github.com/FunAudioLLM/Fun-ASR
• Paper: https://arxiv.org/abs/2305.11013

Would be happy to help with integration if there is interest!