arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2110.03334eess.AS
收藏

从大型自监督预训练模型中提取神经传感器的知识

Knowledge Distillation for Neural Transducers from Large Self-Supervised Pre-trained Models

Xiaoyu Yang, Qiujia Li, Philip C. Woodland

摘要:

自监督预训练是利用大量未标记数据降低自动语音识别(ASR)系统误词率的有效方法。由于对于许多真实世界的ASR应用使用大型预训练模型是不切实际的,因此希望具有小得多的模型,同时保留预训练模型的性能。在本文中,我们提出了一个简单的知识蒸馏(KD)神经传感器的损失函数,专注于流和非流设置下输出概率格中的最佳路径,这使得一个小的学生模型接近大型预训练教师模型的性能。LibriSpeech数据集上的实验表明,尽管比教师模型小10倍,但与使用标记的100小时干净数据在没有KD的情况下训练的基线传感器相比,所提出的损失导致非流和流学生模型的测试其他集的相对WER减少(WERR)为11.5%和6.8%。对于KD额外的860小时未标记数据,WERR增加到48.2%,对于非流和流学生增加到38.5%。如果使用语言模型浅融合来产生蒸馏目标,则观察到学生模型的进一步改进。

英文摘要:

Self-supervised pre-training is an effective approach to leveraging a large amount of unlabelled data to reduce word error rates (WERs) of automatic speech recognition (ASR) systems. Since it is impractical to use large pre-trained models for many real-world ASR applications, it is desirable to have a much smaller model while retaining the performance of the pre-trained model. In this paper, we propose a simple knowledge distillation (KD) loss function for neural transducers that focuses on the one-best path in the output probability lattice under both streaming and non-streaming setups, which allows a small student model to approach the performance of the large pre-trained teacher model. Experiments on the LibriSpeech dataset show that despite being 10 times smaller than the teacher model, the proposed loss results in relative WER reductions (WERRs) of 11.5% and 6.8% on the test-other set for non-streaming and streaming student models compared to the baseline transducers trained without KD using the labelled 100-hour clean data. With an additional 860 hours of unlabelled data for KD, the WERRs increase to 48.2% and 38.5% for non-streaming and streaming students. If language model shallow fusion is used for producing distillation targets, a further improvement in the student model is observed.

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。

↑