Multi-Modal Pre-Training for Automated Speech Recognition

TL;DR

This paper introduces a multi-modal pre-training approach that incorporates environmental context into speech recognition, improving robustness and accuracy over traditional local-only methods.

Contribution

It presents a novel self-supervised multi-modal encoding and deep-fusion framework that enhances ASR performance by leveraging environmental information.

Findings

Up to 7% improvement on Librispeech

6% to 45% gains on internal datasets

Enhanced robustness to noise and corruption

Abstract

Traditionally, research in automated speech recognition has focused on local-first encoding of audio representations to predict the spoken phonemes in an utterance. Unfortunately, approaches relying on such hyper-local information tend to be vulnerable to both local-level corruption (such as audio-frame drops, or loud noises) and global-level noise (such as environmental noise, or background noise) that has not been seen during training. In this work, we introduce a novel approach which leverages a self-supervised learning technique based on masked language modeling to compute a global, multi-modal encoding of the environment in which the utterance occurs. We then use a new deep-fusion framework to integrate this global context into a traditional ASR method, and demonstrate that the resulting method can outperform baseline methods by up to 7% on Librispeech; gains on internal datasets range from 6% (on larger models) to 45% (on smaller models).