Improving Transformer-based Speech Recognition Using Unsupervised Pre-training

TL;DR

This paper introduces an unsupervised pre-training approach called Masked Predictive Coding for Transformer-based speech recognition, significantly improving accuracy with less labeled data.

Contribution

It proposes a novel unsupervised pre-training method for Transformer speech models, enhancing recognition performance without extensive labeled datasets.

Findings

Achieved CER of 23.3% with the same training data, surpassing previous models.

Further reduced CER to 21.0% with additional pre-training data.

Demonstrated 11.8% relative CER reduction over baseline.

Abstract

Speech recognition technologies are gaining enormous popularity in various industrial applications. However, building a good speech recognition system usually requires large amounts of transcribed data, which is expensive to collect. To tackle this problem, an unsupervised pre-training method called Masked Predictive Coding is proposed, which can be applied for unsupervised pre-training with Transformer based model. Experiments on HKUST show that using the same training data, we can achieve CER 23.3%, exceeding the best end-to-end model by over 0.2% absolute CER. With more pre-training data, we can further reduce the CER to 21.0%, or a 11.8% relative CER reduction over baseline.