← demo
PRODUCT · VOICE AGENT

Chinese real-time voice agent

End-to-end STT → LLM → TTS at ~600 ms TTFA, with barge-in and multi-turn context. Pick a scenario below to try.

TTFA
~ 600 ms
BARGE-IN
< 200 ms
LANGUAGES
zh · yue · en
STACK
Aliyun · Qwen
SCENARIOS
TECHNICAL
PIPELINE Aliyun Paraformer STT → Qwen-Max → CosyVoice TTS
HOSTING LiveKit Cloud agent + edge token service
REPO github.com/lake/.../zh-voice-assistant

Want this pipeline running on your own business scenarios?

Book a technical call →