多语言DistilWhisper：通过语言特定专家高效多任务语音模型知识蒸馏
Multilingual DistilWhisper: Efficient Distillation of Multi-task Speech Models via Language-Specific Experts
AI总结：
DistilWhisper通过语言特定专家和知识蒸馏提升欠代表语言的ASR性能，同时保持多任务和多语言能力。
摘要：
Whisper是一个涵盖99种语言的多任务和多语言语音模型。它在部分覆盖语言中能产生优异的自动语音识别（ASR）结果，但在非忽视数量的欠代表语言上仍表现不足，尤其是在较小模型版本中问题更为严重。在本工作中，我们提出了DistilWhisper，一种能够弥合这些语言在ASR上的性能差距同时保留多任务和多语言能力优势的方法。我们的方法包括两个关键策略：使用语言特定专家对Whisper-small进行轻量级模块化ASR微调，以及从Whisper-large-v2进行知识蒸馏。这种双方法使我们能够有效提升ASR性能，同时保留来自多任务和多语言预训练继承的鲁棒性。实验结果表明，我们的方法比标准微调或LoRA适配器更有效，能够在针对语言的in-和out-of-domain测试集上提升性能，同时仅引入微小的参数开销。
英文摘要：
Whisper is a multitask and multilingual speech model covering 99 languages. It yields commendable automatic speech recognition (ASR) results in a subset of its covered languages, but the model still underperforms on a non-negligible number of under-represented languages, a problem exacerbated in smaller model versions. In this work, we propose DistilWhisper, an approach able to bridge the performance gap in ASR for these languages while retaining the advantages of multitask and multilingual capabilities. Our approach involves two key strategies: lightweight modular ASR fine-tuning of whisper-small using language-specific experts, and knowledge distillation from whisper-large-v2. This dual approach allows us to effectively boost ASR performance while keeping the robustness inherited from the multitask and multilingual pre-training. Results demonstrate that our approach is more effective than standard fine-tuning or LoRA adapters, boosting performance in the targeted languages for both in- and out-of-domain test sets, while introducing only a negligible parameter overhead at inference.
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。
