← sources

netease-youdao/Confucius4-R2T2 · Hugging Face

source: reddit (reddit:LocalLLaMA)published 2026-11-09 14:13ingested 2026-11-10 02:42

retained text

# 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. The model operates in a monotonic output mode, committing transcript text permanently without revising previous words. This is critical for applications where text must be processed or acted upon instantly. The result is a smoother user experience, with fewer disruptive text revisions and less visual flicker in real-time applications such as real-time live captioning and subtitling, simultaneous speech translation, downstream NLP pipelines and LLM agents.

R2T2, short for Real Real-Time Transcription, adapts **Qwen3-ASR-1.7B** with **Longest Stable Prefix (LSP) learning** on 1,500 hours of Chinese and English audio. Two-stage training with SFT and GRPO updates only the language decoder; the pre-trained audio encoder and projector remain frozen, with no architectural changes. LSP targets and an explicit `Wait` action teach the model what it can commit and when it needs more audio.

* **GGUF** : [https://huggingface.co/netease-youdao/Confucius4-R2T2-GGUF](https://huggingface.co/netease-youdao/Confucius4-R2T2-GGUF)
* **Space** : [https://huggingface.co/spaces/netease-youdao/confucius4-r2t2-demo](https://huggingface.co/spaces/netease-youdao/confucius4-r2t2-demo)
* **GitHub** : [https://github.com/netease-youdao/Confucius4-R2T2](https://github.com/netease-youdao/Confucius4-R2T2)
* **arXiv** : [https://arxiv.org/abs/2611.06043](https://arxiv.org/abs/2611.06043)
* **PDF** : [https://arxiv.org/pdf/2611.06043](https://arxiv.org/pdf/2611.06043)

# License

R2T2 uses **dual licensing** to distinguish the source code from the model weights:

* **Code** in the accompanying GitHub repository is released under the [Apache License 2.0](https://github.com/netease-youdao/Confucius4-R2T2/blob/master/LICENSE) and is free to use, modify, and redistribute (including commercially) under the terms of that license.
* **Model weights** are released under the [NetEase Model Use License Agreement](https://github.com/netease-youdao/Confucius4-R2T2/blob/master/MODEL_LICENSE).

document fetched 2026-11-10 00:41 · canonical https://reddit.com/r/LocalLLaMA/comments/1xrspiy/neteaseyoudaoconfucius4r2t2_hugging_face