SimulMEGA：MoE 路由器是面向同声传译的高级决策者
SimulMEGA: MoE Routers are Advanced Policy Makers for Simultaneous Speech Translation
AI总结：
针对同声传译中质量、延迟与连贯性难平衡，尤其多语言场景策略学习困难的问题，提出无监督框架SimulMEGA，结合前缀训练与MoE精炼器隐式学习读写决策，在多语言对及流式TTS任务中实现更优延迟-质量权衡。
摘要：
同声传译（SimulST）通过在严格的延迟约束下联合优化语音识别与机器翻译，实现实时跨语言通信。现有系统难以在翻译质量、延迟与语义连贯性之间取得平衡，尤其在多语言多对多场景中，读写策略的差异性阻碍了统一策略的学习。\n本文提出 SimulMEGA（基于混合专家门控的同声生成，Simultaneous Generation by Mixture-of-Experts Gating），这是一种无监督策略学习框架，将基于前缀的训练与混合专家（MoE）精炼器相结合，以隐式方式学习有效的读写决策，且不会增加推理时开销。该设计仅需对标准 Transformer 架构做极小修改，可泛化至语音到文本和文本到语音的流式任务。\n通过在6个语言对上的综合评估，我们5亿参数的语音到文本模型性能优于 Seamless 基线：在平均延迟1.5秒时，BLEU 下降幅度低于7%；在3秒延迟时，下降幅度低于3%。我们进一步将 SimulMEGA 扩展至采用单向骨干网络的流式文本转语音（TTS）任务，验证了其通用性，实现了更优的延迟-质量权衡。
英文摘要：
Simultaneous Speech Translation (SimulST) enables real-time cross-lingual communication by jointly optimizing speech recognition and machine translation under strict latency constraints. Existing systems struggle to balance translation quality, latency, and semantic coherence, particularly in multilingual many-to-many scenarios where divergent read and write policies hinder unified strategy learning. In this paper, we present SimulMEGA (Simultaneous Generation by Mixture-of-Experts Gating), an unsupervised policy learning framework that combines prefix-based training with a Mixture-of-Experts refiner to learn effective read and write decisions in an implicit manner, without adding inference-time overhead. Our design requires only minimal modifications to standard transformer architectures and generalizes across both speech-to-text and text-to-speech streaming tasks. Through comprehensive evaluation on six language pairs, our 500M parameter speech-to-text model outperforms the Seamless baseline, achieving under 7 percent BLEU degradation at 1.5 seconds average lag and under 3 percent at 3 seconds. We further demonstrate the versatility of SimulMEGA by extending it to streaming TTS with a unidirectional backbone, yielding superior latency quality tradeoffs.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
