语音识别不再改口
实时语音识别有个隐蔽的毛病：它边说边改。屏幕上先蹦出「今天天汽不错」，半秒后悄悄改成「天气」;对看字幕的人只是闪一下，对机器就是灾难——翻译器已经把错词译出去了，语音助手已经照错指令动了手。这篇的做法是让模型「说出口的话永不收回」:先定义什么叫「最长稳定前缀」，即当前输出里可以确定不会再变的那一段，并证明它只会变长、不会缩回，且与听完整个音频后的结果一致；再用这种标注去后训练一个现成的中文语音模型，只训 1500 小时中英文音频、冻结原有听觉编码器，不动架构。结果是流式识别的准确率接近「听完再转」的水平，延迟也低。做实时字幕、同声传译、语音助手的人，这是直接对症的一步：错一次就执行错的机器，比慢半秒的机器贵得多。
📄 原文摘要(英文)
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.