Confucius4-R2T2: A Stable-Prefix Streaming ASR Model for Real Real-Time Transcription

Dani HG, Qi He, Shaoxiong Duan, Yingjie Li, chengqiao, Yitao Duan

Confucius4-R2T2: A Stable-Prefix Streaming ASR Model for Real Real-Time Transcription: 92 upvotes on Hugging Face Daily Papers, #1 of 34 papers on 2026-11-09. Day-by-day upvote history.

Upvotes over time

It had 26 upvotes by its first day, 90 upvotes by the next day, 92 upvotes by day 3, 92 upvotes by day 7, and has 92 now. It ranked #2 of 133 papers in its week and #7 of 340 in its month. Its code repository has 885 GitHub stars.

Streaming speech recognition is increasingly consumed by machines, such as translators and voice agents, that act on each token as it appears. A token that is later revised costs wasted work or a wrong action. In this work, we present Confucius4-R2T2 (short for Real Real-Time Transcription), a stable-prefix streaming ASR model that never modifies its earlier output while maintaining high accuracy and low latency. R2T2 is obtained by post-training Qwen3-ASR-1.7B on 1,500 hours of Chinese and English audio while freezing its pre-trained audio encoder and projector. The adaptation is based on the concept of Longest Stable Prefix (LSP), a new notion that we introduce in this work. We prove that LSP is monotone under the prefix order, consistent with the full-audio output, and maximal among labels that equal a whole earlier output. Learning from LSP labels converts a full-context model into a stable-prefix streaming one without any architectural change. Extensive evaluations show that R2T2 brings streaming recognition close to the base model's full-context quality while delivering stable text with low emission lag. Its gains extend beyond the post-training languages, demonstrating a practical path from full-context ASR to stable streaming. We open-source the model weights and code at https://huggingface.co/netease-youdao/Confucius4-R2T2 and https://github.com/netease-youdao/Confucius4-R2T2.

Paper page on Hugging Face · arXiv

Data: hysts-bot-data/daily-papers-stats and the Daily Papers API. Open data: tardellirs/paper-pulse-data. Sister project: Model Pulse, the download history of every model on the Hub.