Feature Request: Add FunASR as speech recognition backend

[LauraGPT]

Note

License and capability clarification (2026-07-14): FunASR is a toolkit, not a single checkpoint. The FunASR and SenseVoice repository source code is MIT; model weights follow each model card. SenseVoiceSmall supports Chinese, Cantonese, English, Japanese, and Korean, and its weights use the linked FunASR Model Open Source License Agreement. Fun-ASR-Nano-2512 is Apache-2.0. Language coverage, punctuation, and performance depend on the selected model and runtime configuration.

Hi! Ichigo is an interesting local realtime voice AI project.

I'd like to suggest adding FunASR (16K+ stars) as an ASR backend option:

• 170x real-time GPU speed — minimal latency for real-time voice interaction
• Streaming ASR: Paraformer-streaming model designed for real-time with sub-second latency
• Built-in VAD + punctuation: No separate preprocessing needed
• 50+ languages: SenseVoice model with automatic language detection
• Local/offline: Runs entirely locally, aligning with Ichigo's local-first approach
• OpenAI-compatible API: funasr-server --device cuda serves at /v1/audio/transcriptions

from funasr import AutoModel
model = AutoModel(model="iic/SenseVoiceSmall")
result = model.generate(input="audio.wav")

Happy to help with integration!