arXiv cs.CL· Haotian Tan, Hiroki Ouchi, Sakriani Sakti·· 14 小时前AI 评分29
SimPOTrans：用强化学习让离线语音翻译模型自适应同传
SimPOTrans: Self-Adaptation for Offline-to-Simultaneous Speech Translation via Reinforcement Learning
AI 导读
SimPOTrans 是一个强化学习框架，通过两阶段 RL 后训练让离线语音翻译模型自适应对话式同传，无需合成读写轨迹或显式延迟优化。推理时无需外部决策策略，仅调整语音输入 chunk size 即可灵活控制质量与延迟的权衡。在 ACL 60/60 和 MCIF 基准的 En→De、En→Zh、En→Ja 三个方向上，该方法优于多个强基线，取得更优的质量—延迟权衡。
正文 · 原文
Abstract:Simultaneous speech translation (SimulST) aims to translate ongoing speech with low latency while preserving translation quality. Recent studies have adopted large language models (LLMs) for SimulST and formulated it as a multi-turn dialogue task for efficient generation, achieving state-of-the-art performance. However, existing dialogue-style SimulST approaches typically rely on specialized training schemes to acquire simultaneous translation behavior, either using synthesized read/write trajectories, which may provide suboptimal training signals, or jointly optimizing translation quality and latency, whose competing objectives may cause optimization instability. In this paper, we introduce SimPOTrans, a reinforcement learning (RL) framework that enables offline speech translation models to self-adapt to dialogue-style SimulST. Through a simple two-stage RL post-training pipeline, SimPOTrans rewards an offline model for exhibiting translation behaviors beneficial to simultaneous scenarios when given partial speech inputs, thereby enabling offline-to-simultaneous self-adaptation without synthesized read/write trajectories or explicit latency optimization. At inference time, SimPOTrans requires no external decision policy and enables flexible quality--latency control by simply adjusting the chunk size of speech input. Experiments on the ACL 60/60 and MCIF benchmarks across three language directions (En$\rightarrow$De, En$\rightarrow$Zh, and En$\rightarrow$Ja) demonstrate that SimPOTrans outperforms several strong baselines, achieving a superior quality--latency trade-off.
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2610.35042 [cs.CL] |
| (or arXiv:2610.35042v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.35042 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Haotian Tan [view email]
[v1]
Mon, 26 Oct 2026 08:54:00 UTC (181 KB)
来源：arXiv cs.CL · arxiv.org