Wav2vec-S: Semi-Supervised Pre-Training for Low-Resource ASR

TL;DR

Wav2vec-S introduces a task-specific semi-supervised pre-training method that refines self-supervised models for low-resource ASR, significantly improving performance with minimal additional pre-training time.

Contribution

The paper proposes wav2vec-S, a semi-supervised pre-training approach that enhances self-supervised models specifically for low-resource ASR tasks, outperforming previous methods.

Findings

Significant WER reductions on various datasets.

Minimal increase in pre-training time.

Semi-supervised pre-training closes representation gaps.

Abstract

Self-supervised pre-training could effectively improve the performance of low-resource automatic speech recognition (ASR). However, existing self-supervised pre-training are task-agnostic, i.e., could be applied to various downstream tasks. Although it enlarges the scope of its application, the capacity of the pre-trained model is not fully utilized for the ASR task, and the learned representations may not be optimal for ASR. In this work, in order to build a better pre-trained model for low-resource ASR, we propose a pre-training approach called wav2vec-S, where we use task-specific semi-supervised pre-training to refine the self-supervised pre-trained model for the ASR task thus more effectively utilize the capacity of the pre-trained model to generate task-specific representations for ASR. Experiments show that compared to wav2vec 2.0, wav2vec-S only requires a marginal increment of pre-training time but could significantly improve ASR performance on in-domain, cross-domain and cross-lingual datasets. Average relative WER reductions are 24.5% and 6.6% for 1h and 10h fine-tuning, respectively. Furthermore, we show that semi-supervised pre-training could close the representation gap between the self-supervised pre-trained model and the corresponding fine-tuned model through canonical correlation analysis.