Reducing Geographic Disparities in Automatic Speech Recognition via Elastic Weight Consolidation

TL;DR

This paper introduces an elastic weight consolidation method to reduce geographic performance disparities in automatic speech recognition, improving accuracy in high-error regions without sacrificing overall performance.

Contribution

The paper proposes using EWC regularization to adapt ASR models for high-error regions while preserving overall accuracy, addressing geographic disparities in ASR.

Findings

WER in high-error regions reduced by 3.2% relative

Overall WER decreased by 1.3% relative

EWC effectively balances regional and overall ASR performance

Abstract

We present an approach to reduce the performance disparity between geographic regions without degrading performance on the overall user population for ASR. A popular approach is to fine-tune the model with data from regions where the ASR model has a higher word error rate (WER). However, when the ASR model is adapted to get better performance on these high-WER regions, its parameters wander from the previous optimal values, which can lead to worse performance in other regions. In our proposed method, we utilize the elastic weight consolidation (EWC) regularization loss to identify directions in parameters space along which the ASR weights can vary to improve for high-error regions, while still maintaining performance on the speaker population overall. Our results demonstrate that EWC can reduce the word error rate (WER) in the region with highest WER by 3.2% relative while reducing the overall WER by 1.3% relative. We also evaluate the role of language and acoustic models in ASR fairness and propose a clustering algorithm to identify WER disparities based on geographic region.