Multilingual Zero Resource Speech Recognition Base on Self-Supervise Pre-Trained Acoustic Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wang, Haoyu, Zhang, Wei-Qiang, Suo, Hongbin, Wan, Yulong
Format: Preprint
Published: 2022
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910779796619264
author Wang, Haoyu
Zhang, Wei-Qiang
Suo, Hongbin
Wan, Yulong
author_facet Wang, Haoyu
Zhang, Wei-Qiang
Suo, Hongbin
Wan, Yulong
contents Labeled audio data is insufficient to build satisfying speech recognition systems for most of the languages in the world. There have been some zero-resource methods trying to perform phoneme or word-level speech recognition without labeled audio data of the target language, but the error rate of these methods is usually too high to be applied in real-world scenarios. Recently, the representation ability of self-supervise pre-trained models has been found to be extremely beneficial in zero-resource phoneme recognition. As far as we are concerned, this paper is the first attempt to extend the use of pre-trained models into word-level zero-resource speech recognition. This is done by fine-tuning the pre-trained models on IPA phoneme transcriptions and decoding with a language model trained on extra texts. Experiments on Wav2vec 2.0 and HuBERT models show that this method can achieve less than 20% word error rate on some languages, and the average error rate on 8 languages is 33.77%.
format Preprint
id arxiv_https___arxiv_org_abs_2210_06936
institution arXiv
publishDate 2022
record_format arxiv
spellingShingle Multilingual Zero Resource Speech Recognition Base on Self-Supervise Pre-Trained Acoustic Models
Wang, Haoyu
Zhang, Wei-Qiang
Suo, Hongbin
Wan, Yulong
Computation and Language
Sound
Audio and Speech Processing
Labeled audio data is insufficient to build satisfying speech recognition systems for most of the languages in the world. There have been some zero-resource methods trying to perform phoneme or word-level speech recognition without labeled audio data of the target language, but the error rate of these methods is usually too high to be applied in real-world scenarios. Recently, the representation ability of self-supervise pre-trained models has been found to be extremely beneficial in zero-resource phoneme recognition. As far as we are concerned, this paper is the first attempt to extend the use of pre-trained models into word-level zero-resource speech recognition. This is done by fine-tuning the pre-trained models on IPA phoneme transcriptions and decoding with a language model trained on extra texts. Experiments on Wav2vec 2.0 and HuBERT models show that this method can achieve less than 20% word error rate on some languages, and the average error rate on 8 languages is 33.77%.
title Multilingual Zero Resource Speech Recognition Base on Self-Supervise Pre-Trained Acoustic Models
topic Computation and Language
Sound
Audio and Speech Processing
url https://arxiv.org/abs/2210.06936