Feature Request: Add FunASR/SenseVoice for on-device ASR

Feature Request

FunASR models are available for on-device deployment via Sherpa-ONNX, which would be ideal for Duix-Mobile's real-time interactive AI avatar.

Why relevant for mobile digital humans?

• Sherpa-ONNX: Runs SenseVoice and Paraformer models on iOS, Android, and Raspberry Pi — no server needed
• SenseVoice: Ultra-fast ASR (234M params) with emotion detection — perfect for driving avatar expressions
• Low latency: Non-autoregressive architecture means instant results, critical for real-time interaction
• 50+ languages: Broad multilingual support for international deployment
• Privacy: Fully on-device, no data leaves the phone

On-device integration via Sherpa-ONNX:

// Already supports iOS, Android, Flutter, React Native
// See: https://github.com/k2-fsa/sherpa-onnx

• GitHub: https://github.com/modelscope/FunASR (16K+ stars)
• SenseVoice: https://github.com/FunAudioLLM/SenseVoice (8K+ stars)
• Sherpa-ONNX: https://github.com/k2-fsa/sherpa-onnx (5K+ stars)