TY - GEN
T1 - Residual Adapters for Parameter-Efficient ASR Adaptation to Atypical and Accented Speech
AU - Tomanek, Katrin
AU - Zayats, Vicky
AU - Padfield, Dirk
AU - Vaillancourt, Kara
AU - Biadsy, Fadi
N1 - Publisher Copyright:
© 2021 Association for Computational Linguistics
PY - 2021/1/1
Y1 - 2021/1/1
N2 - Automatic Speech Recognition (ASR) systems are often optimized to work best for speakers with canonical speech patterns. Unfortunately, these systems perform poorly when tested on atypical speech and heavily accented speech. It has previously been shown that personalization through model fine-tuning substantially improves performance. However, maintaining such large models per speaker is costly and difficult to scale. We show that by adding a relatively small number of extra parameters to the encoder layers via so-called residual adapter, we can achieve similar adaptation gains compared to model fine-tuning, while only updating a tiny fraction (less than 0.5%) of the model parameters. We demonstrate this on two speech adaptation tasks (atypical and accented speech) and for two state-of-the-art ASR architectures.
AB - Automatic Speech Recognition (ASR) systems are often optimized to work best for speakers with canonical speech patterns. Unfortunately, these systems perform poorly when tested on atypical speech and heavily accented speech. It has previously been shown that personalization through model fine-tuning substantially improves performance. However, maintaining such large models per speaker is costly and difficult to scale. We show that by adding a relatively small number of extra parameters to the encoder layers via so-called residual adapter, we can achieve similar adaptation gains compared to model fine-tuning, while only updating a tiny fraction (less than 0.5%) of the model parameters. We demonstrate this on two speech adaptation tasks (atypical and accented speech) and for two state-of-the-art ASR architectures.
UR - https://www.scopus.com/pages/publications/85127411161
U2 - 10.18653/v1/2021.emnlp-main.541
DO - 10.18653/v1/2021.emnlp-main.541
M3 - Conference contribution
AN - SCOPUS:85127411161
T3 - EMNLP 2021 - 2021 Conference on Empirical Methods in Natural Language Processing, Proceedings
SP - 6751
EP - 6760
BT - EMNLP 2021 - 2021 Conference on Empirical Methods in Natural Language Processing, Proceedings
PB - Association for Computational Linguistics (ACL)
T2 - 2021 Conference on Empirical Methods in Natural Language Processing, EMNLP 2021
Y2 - 7 November 2021 through 11 November 2021
ER -