BriefGPT.xyz
Jun, 2022
Wav2Vec-Aug: 有限数据下的改进自监督训练
Wav2Vec-Aug: Improved self-supervised training with limited data
HTML
PDF
Anuroop Sriram, Michael Auli, Alexei Baevski
TL;DR
本文探讨了如何利用数据增强技术来解决语言和领域中数据不足的问题，从而在限制条件下将自监督学习应用于语音表示方面，并在Wav2Vec 2.0预训练模型的每个组成部分上提出了改进方法，在Librispeech的测试任务中比Wav2Vec 2.0实现了13％的相对字错误率改进。
Abstract
self-supervised learning
(SSL) of
speech representations
has received much attention over the last few years but most work has focused on languages and domains with an abundance of unlabeled data. However, for ma
→