PROCTER: PROnunciation-aware ConTextual adaptER for personalized speech recognition in neural transducers

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Pandey, Rahul, Ren, Roger, Luo, Qi, Liu, Jing, Rastrow, Ariya, Gandhe, Ankur, Filimonov, Denis, Strimel, Grant, Stolcke, Andreas, Bulyko, Ivan
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929237349367808
author Pandey, Rahul
Ren, Roger
Luo, Qi
Liu, Jing
Rastrow, Ariya
Gandhe, Ankur
Filimonov, Denis
Strimel, Grant
Stolcke, Andreas
Bulyko, Ivan
author_facet Pandey, Rahul
Ren, Roger
Luo, Qi
Liu, Jing
Rastrow, Ariya
Gandhe, Ankur
Filimonov, Denis
Strimel, Grant
Stolcke, Andreas
Bulyko, Ivan
contents End-to-End (E2E) automatic speech recognition (ASR) systems used in voice assistants often have difficulties recognizing infrequent words personalized to the user, such as names and places. Rare words often have non-trivial pronunciations, and in such cases, human knowledge in the form of a pronunciation lexicon can be useful. We propose a PROnunCiation-aware conTextual adaptER (PROCTER) that dynamically injects lexicon knowledge into an RNN-T model by adding a phonemic embedding along with a textual embedding. The experimental results show that the proposed PROCTER architecture outperforms the baseline RNN-T model by improving the word error rate (WER) by 44% and 57% when measured on personalized entities and personalized rare entities, respectively, while increasing the model size (number of trainable parameters) by only 1%. Furthermore, when evaluated in a zero-shot setting to recognize personalized device names, we observe 7% WER improvement with PROCTER, as compared to only 1% WER improvement with text-only contextual attention
format Preprint
id arxiv_https___arxiv_org_abs_2303_17131
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle PROCTER: PROnunciation-aware ConTextual adaptER for personalized speech recognition in neural transducers
Pandey, Rahul
Ren, Roger
Luo, Qi
Liu, Jing
Rastrow, Ariya
Gandhe, Ankur
Filimonov, Denis
Strimel, Grant
Stolcke, Andreas
Bulyko, Ivan
Audio and Speech Processing
Sound
End-to-End (E2E) automatic speech recognition (ASR) systems used in voice assistants often have difficulties recognizing infrequent words personalized to the user, such as names and places. Rare words often have non-trivial pronunciations, and in such cases, human knowledge in the form of a pronunciation lexicon can be useful. We propose a PROnunCiation-aware conTextual adaptER (PROCTER) that dynamically injects lexicon knowledge into an RNN-T model by adding a phonemic embedding along with a textual embedding. The experimental results show that the proposed PROCTER architecture outperforms the baseline RNN-T model by improving the word error rate (WER) by 44% and 57% when measured on personalized entities and personalized rare entities, respectively, while increasing the model size (number of trainable parameters) by only 1%. Furthermore, when evaluated in a zero-shot setting to recognize personalized device names, we observe 7% WER improvement with PROCTER, as compared to only 1% WER improvement with text-only contextual attention
title PROCTER: PROnunciation-aware ConTextual adaptER for personalized speech recognition in neural transducers
topic Audio and Speech Processing
Sound
url https://arxiv.org/abs/2303.17131