Correcting Automated and Manual Speech Transcription Errors using Warped Language Models

TL;DR

This paper introduces a novel approach using warped language models to effectively correct errors in automatic and manual speech transcriptions, significantly reducing word error rates.

Contribution

The work presents a new method leveraging warped language models' robustness to transcription noise for improved speech transcription correction.

Findings

Achieves up to 10% reduction in word error rates

Effective on both automatic and manual transcriptions

Demonstrates robustness to different types of transcription errors

Abstract

Masked language models have revolutionized natural language processing systems in the past few years. A recently introduced generalization of masked language models called warped language models are trained to be more robust to the types of errors that appear in automatic or manual transcriptions of spoken language by exposing the language model to the same types of errors during training. In this work we propose a novel approach that takes advantage of the robustness of warped language models to transcription noise for correcting transcriptions of spoken language. We show that our proposed approach is able to achieve up to 10% reduction in word error rates of both automatic and manual transcriptions of spoken language.