Feature Request: Add SenseVoice for on-device voice recognition

[LauraGPT]

Feature Request

Owl is a great personal wearable AI that runs locally. Suggesting SenseVoice as an ASR option — its compact size and speed make it ideal for local/edge deployment.

Why SenseVoice for wearable AI?

• 234M params — compact enough for local inference on consumer hardware
• Non-autoregressive — constant low latency (~100ms), no waiting for sequential decoding
• 5x faster than Whisper — better battery life and responsiveness
• 50+ languages — single model handles any language
• Emotion detection — contextual awareness of speaker mood
• Audio events — detects environmental sounds (laughter, music, etc.)

Local deployment options

1. Python (simplest): pip install funasr
2. ONNX (optimized): via Sherpa-ONNX — supports ARM, x86, mobile
3. C++ (embedded): SenseVoice.cpp — no Python dependency

Quick start

from funasr import AutoModel
model = AutoModel(model="iic/SenseVoiceSmall", vad_model="fsmn-vad")
result = model.generate(input=audio)

• FunASR: https://github.com/modelscope/FunASR (16.6K stars)
• SenseVoice: https://github.com/FunAudioLLM/SenseVoice (8.3K stars)