仅更新1%-5%参数，高效适配低资源新语言的语音识别模型。
How to Learn a New Language? An Efficient Solution for Self-Supervised Learning Models Unseen Languages Adaption in Low-Resource Scenario
- 引入中间适应阶段，优化适配器初始化，提升训练效率。
- 在未见语言上实现字符/音素错误率降低28%的相对提升。
- 适合资源匮乏场景下快速部署新语言语音识别系统。
自监督语音学习（SSL）模型在自动语音识别（ASR）中表现优异，但在低资源语言场景下面临预训练与目标语言间的领域差异问题。传统微调计算开销大，而冻结SSL模型作为特征提取器则性能不佳。为此，本文基于适配器提出一种高效微调方案，增加中间适应阶段以优化适配器和下游模型的初始状态。实验表明，仅更新总参数的1%-5%，即可在ML-SUPERB数据集上实现对未见语言的有效适配，字符/音素错误率相对降低高达28%。
原文摘要 · Abstract (English)
The utilization of speech Self-Supervised Learning (SSL) models achieves impressive performance on Automatic Speech Recognition (ASR). However, in low-resource language ASR, they encounter the domain mismatch problem between pre-trained and low-resource languages. Typical solutions like fine-tuning the SSL model suffer from high computation costs while using frozen SSL models as feature extractors comes with poor performance. To handle these issues, we extend a conventional efficient fine-tuning scheme based on the adapter. We add an extra intermediate adaptation to warm up the adapter and downstream model initialization. Remarkably, we update only 1-5% of the total model parameters to achieve the adaptation. Experimental results on the ML-SUPERB dataset show that our solution outperforms conventional efficient fine-tuning. It achieves up to a 28% relative improvement in the Character/Phoneme error rate when adapting to unseen languages.
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。