netease-youdao/Confucius4-R2T2 · Hugging Face
Confucius4-R2T2: A Stable-Prefix Streaming ASR Model for Real Real-Time Transcription Confucius4-R2T2 is a low-latency, high-accuracy stable-prefix streaming automatic speech recognition (ASR) model. It supports configurable decoding chunks from 80 ms to 2 s.