Pushing the performances of ASR models on English and Spanish accents
This work addresses the issue of accent bias in ASR systems for users with diverse English and Spanish accents, but it is incremental as it builds on existing methods.
The paper tackled the problem of automatic speech recognition (ASR) models performing poorly on non-standard accents by applying pre-trained embeddings and auxiliary classification losses, resulting in improved performance across multiple model architectures and languages.
Speech to text models tend to be trained and evaluated against a single target accent. This is especially true for English for which native speakers from the United States became the main benchmark. In this work, we are going to show how two simple methods: pre-trained embeddings and auxiliary classification losses can improve the performance of ASR systems. We are looking for upgrades as universal as possible and therefore we will explore their impact on several models architectures and several languages.