arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2202.09167eess.AScs.SD
收藏

使用训练良好的ASR Conformer模型对低资源目标域模型进行领域自适应

Domain Adaptation of low-resource Target-Domain models using well-trained ASR Conformer Models

IIT Madras(印度理工学院马德拉斯分校)

Vrunda N. Sukhadia, S. Umesh

AI总结:

本文提出利用训练良好的ASR模型编码器层嵌入作为低资源目标域Conformer模型的特征,并辅以SpecAug增强,在LibriSpeech-100-clean和WSJ上分别取得30%和50%的相对改进。

摘要:

在本文中,我们研究了当存在一个使用大数据集训练的良好自动语音识别(ASR)模型时,针对低资源目标域数据的领域自适应问题。我们认为,在编码器-解码器框架中,训练良好的ASR模型的解码器在很大程度上偏向源域,从而在普通的迁移学习中损害了目标域模型的性能。另一方面,训练良好的ASR模型的编码器层主要捕获声学特征。因此,我们提出将从这些编码器层提取的嵌入作为下游Conformer目标域模型的特征,并表明它们能带来显著的改进。我们进行了消融研究,以确定从哪个编码器层提取嵌入是最优的,以及冻结或更新训练良好的ASR模型编码器层的效果。我们进一步表明,在所提出的特征上应用频谱增强(SpecAug)(这是在输入频谱特征上默认应用SpecAug之外的额外操作)能进一步改善目标域性能。对于以LibriSpeech-100-clean数据作为目标域、SPGI-5000作为训练良好模型的情况,我们相对于基线获得了30%的相对改进。同样,以WSJ数据作为目标域、LibriSpeech-960作为训练良好模型,我们相对于基线获得了50%的相对改进。

英文摘要:

In this paper, we investigate domain adaptation for low-resource Automatic Speech Recognition (ASR) of target-domain data, when a well-trained ASR model trained with a large dataset is available. We argue that in the encoder-decoder framework, the decoder of the well-trained ASR model is largely tuned towards the source-domain, hurting the performance of target-domain models in vanilla transfer-learning. On the other hand, the encoder layers of the well-trained ASR model mostly capture the acoustic characteristics. We, therefore, propose to use the embeddings tapped from these encoder layers as features for a downstream Conformer target-domain model and show that they provide significant improvements. We do ablation studies on which encoder layer is optimal to tap the embeddings, as well as the effect of freezing or updating the well-trained ASR model's encoder layers. We further show that applying Spectral Augmentation (SpecAug) on the proposed features (this is in addition to default SpecAug on input spectral features) provides a further improvement on the target-domain performance. For the LibriSpeech-100-clean data as target-domain and SPGI-5000 as a well-trained model, we get 30% relative improvement over baseline. Similarly, with WSJ data as target-domain and LibriSpeech-960 as a well-trained model, we get 50% relative improvement over baseline.

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。

↑