BERT Meets CTC: New Formulation of End-to-End Speech Recognition with Pre-trained Masked Language Model

TL;DR

This paper introduces BERT-CTC, a new end-to-end speech recognition model that integrates BERT's contextual embeddings with CTC, relaxing independence assumptions and enhancing linguistic understanding for improved accuracy and downstream task performance.

Contribution

It proposes a novel formulation combining BERT with CTC for speech recognition, enabling contextual dependencies and improving robustness across languages and speaking styles.

Findings

Outperforms conventional CTC-based models in accuracy

Enhances robustness across languages and speaking styles

Benefits downstream spoken language understanding tasks

Abstract

This paper presents BERT-CTC, a novel formulation of end-to-end speech recognition that adapts BERT for connectionist temporal classification (CTC). Our formulation relaxes the conditional independence assumptions used in conventional CTC and incorporates linguistic knowledge through the explicit output dependency obtained by BERT contextual embedding. BERT-CTC attends to the full contexts of the input and hypothesized output sequences via the self-attention mechanism. This mechanism encourages a model to learn inner/inter-dependencies between the audio and token representations while maintaining CTC's training efficiency. During inference, BERT-CTC combines a mask-predict algorithm with CTC decoding, which iteratively refines an output sequence. The experimental results reveal that BERT-CTC improves over conventional approaches across variations in speaking styles and languages. Finally, we show that the semantic representations in BERT-CTC are beneficial towards downstream spoken language understanding tasks.