GMU改进了低资源语音翻译模型，提升多语言翻译效果。
GMU Systems for the IWSLT 2025 Low-Resource Speech Translation Shared Task
- 用SeamlessM4T-v2微调端到端语音翻译，支持多语言
- 直接微调在多数语言上表现最佳，初始化编码器提升新语言性能
- 多任务训练略有帮助，适合低资源场景研究者
本文介绍GMU团队参与IWSLT 2025低资源语音翻译共享任务的系统。我们为除黎凡特阿拉伯语外的所有语言对训练了系统，基于SeamlessM4T-v2微调自动语音识别（ASR）、机器翻译（MT）和端到端语音翻译（E2E ST）模型。其中ASR与MT模型也用于构建级联式语音翻译系统。此外，我们探索了多种端到端训练范式，包括直接微调、多任务训练以及使用微调过的ASR和/或MT模型进行参数初始化。实验结果表明：（1）直接端到端微调能取得良好效果；（2）使用已微调的ASR编码器初始化可提升未在SeamlessM4T-v2上训练过的语言的翻译性能；（3）多任务训练有轻微增益。
原文摘要 · Abstract (English)
This paper describes the GMU systems for the IWSLT 2025 low-resource speech translation shared task. We trained systems for all language pairs, except for Levantine Arabic. We fine-tuned SeamlessM4T-v2 for automatic speech recognition (ASR), machine translation (MT), and end-to-end speech translation (E2E ST). The ASR and MT models are also used to form cascaded ST systems. Additionally, we explored various training paradigms for E2E ST fine-tuning, including direct E2E fine-tuning, multi-task training, and parameter initialization using components from fine-tuned ASR and/or MT models. Our results show that (1) direct E2E fine-tuning yields strong results; (2) initializing with a fine-tuned ASR encoder improves ST performance on languages SeamlessM4T-v2 has not been trained on; (3) multi-task training can be slightly helpful.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。