NetEase Youdao's Confucius4-R2T2: stable-prefix streaming ASR that never revises its output
netease-youdao · hf · 2026-11-09
NetEase Youdao released Confucius4-R2T2, a stable-prefix streaming ASR model whose output is never revised while maintaining near full-context accuracy and low latency. It post-trains Qwen3-ASR-1.7B on 1,500 hours of Chinese and English audio with the encoder and projector frozen, using their new Longest Stable Prefix (LSP) concept—proven monotone under the prefix order and consistent with full-audio output. Aimed at machine consumers like translators and voice agents that act on every token as it appears.
More from Multimodal
- EvoWorld Teaches Video Generators Causal World Evolution Beyond the Frame — Yequan Bie · 2026-11-09
- AI-generated music nails odd time signatures, key changes and 8-bit vibes — sebkrier · 2026-11-09
- Band playing cardboard instruments: a nostalgic fail from old ModelScope video gen — MiverOfficial · 2026-11-09
- AI-generated retro Otaru snow night video paired with original 90s-style J-R&B track — Sumica7 · 2026-11-09
- User asks GPT 6.2 to compose a symphony and shares the result — imjustnewatai · 2026-11-09
- AI image creators show off that one 'perfect generation' — PurzBeats · 2026-11-09