Improving Accented Speech Recognition with Multi-Domain Training

2023/03/14 by Maison, Lucas, Estève, Yannick · 2 citations
#Audio and Speech Processing (eess.AS) #Computation and Language (cs.CL) #FOS: Computer and information sciences #FOS: Electrical engineering #Machine Learning (cs.LG) #Sound (cs.SD) #electronic engineering #information engineering

paper · doi:10.48550/arxiv.2303.07924

Abstract

Thanks to the rise of self-supervised learning, automatic speech recognition (ASR) systems now achieve near-human performance on a wide variety of datasets. However, they still lack generalization capability and are not robust to domain shifts like accent variations. In this work, we use speech audio representing four different French accents to create fine-tuning datasets that improve the robustness of pre-trained ASR models. By incorporating various accents in the training set, we obtain both in-domain and out-of-domain improvements. Our numerical experiments show that we can reduce error rates by up to 25% (relative) on African and Belgian accents compared to single-domain training while keeping a good performance on standard French.

Cited by

• Personal Attribute Leakage in Federated Speech Models
• Benchmarking Large Pretrained Multilingual Models on Québec French Speech Recognition

Related

• Improving Accent Identification and Accented Speech Recognition Under a Framework of Self-supervised Learning
• Analysis of French Phonetic Idiosyncrasies for Accent Recognition
• Accented Speech Recognition With Accent-specific Codebooks
• Accented Speech Recognition: A Survey
• Best of Both Worlds: Robust Accented Speech Recognition with Adversarial\n Transfer Learning