Speech Recognition with Augmented Synthesized Speech

TL;DR

This paper investigates using synthesized speech generated by Tacotron for data augmentation to improve speech recognition across domains, showing potential but highlighting existing performance gaps.

Contribution

It evaluates the effectiveness of synthesized speech for data augmentation in speech recognition and explores algorithms to enhance acoustic and lexical diversity.

Findings

Synthesized speech can improve recognition accuracy when used for data augmentation.

There is a significant performance gap between recognizers trained on human versus synthesized speech.

Augmentation with synthesized speech shows promise for domain transfer in speech recognition.

Abstract

Recent success of the Tacotron speech synthesis architecture and its variants in producing natural sounding multi-speaker synthesized speech has raised the exciting possibility of replacing expensive, manually transcribed, domain-specific, human speech that is used to train speech recognizers. The multi-speaker speech synthesis architecture can learn latent embedding spaces of prosody, speaker and style variations derived from input acoustic representations thereby allowing for manipulation of the synthesized speech. In this paper, we evaluate the feasibility of enhancing speech recognition performance using speech synthesis using two corpora from different domains. We explore algorithms to provide the necessary acoustic and lexical diversity needed for robust speech recognition. Finally, we demonstrate the feasibility of this approach as a data augmentation strategy for domain-transfer. We find that improvements to speech recognition performance is achievable by augmenting training data with synthesized material. However, there remains a substantial gap in performance between recognizers trained on human speech those trained on synthesized speech.