CR-CTC: Consistency regularization on CTC for improved speech recognition

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Yao, Zengwei, Kang, Wei, Yang, Xiaoyu, Kuang, Fangjun, Guo, Liyong, Zhu, Han, Jin, Zengrui, Li, Zhaoqing, Lin, Long, Povey, Daniel
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913690137133056
author Yao, Zengwei
Kang, Wei
Yang, Xiaoyu
Kuang, Fangjun
Guo, Liyong
Zhu, Han
Jin, Zengrui
Li, Zhaoqing
Lin, Long
Povey, Daniel
author_facet Yao, Zengwei
Kang, Wei
Yang, Xiaoyu
Kuang, Fangjun
Guo, Liyong
Zhu, Han
Jin, Zengrui
Li, Zhaoqing
Lin, Long
Povey, Daniel
contents Connectionist Temporal Classification (CTC) is a widely used method for automatic speech recognition (ASR), renowned for its simplicity and computational efficiency. However, it often falls short in recognition performance. In this work, we propose the Consistency-Regularized CTC (CR-CTC), which enforces consistency between two CTC distributions obtained from different augmented views of the input speech mel-spectrogram. We provide in-depth insights into its essential behaviors from three perspectives: 1) it conducts self-distillation between random pairs of sub-models that process different augmented views; 2) it learns contextual representation through masked prediction for positions within time-masked regions, especially when we increase the amount of time masking; 3) it suppresses the extremely peaky CTC distributions, thereby reducing overfitting and improving the generalization ability. Extensive experiments on LibriSpeech, Aishell-1, and GigaSpeech datasets demonstrate the effectiveness of our CR-CTC. It significantly improves the CTC performance, achieving state-of-the-art results comparable to those attained by transducer or systems combining CTC and attention-based encoder-decoder (CTC/AED). We release our code at https://github.com/k2-fsa/icefall.
format Preprint
id arxiv_https___arxiv_org_abs_2410_05101
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle CR-CTC: Consistency regularization on CTC for improved speech recognition
Yao, Zengwei
Kang, Wei
Yang, Xiaoyu
Kuang, Fangjun
Guo, Liyong
Zhu, Han
Jin, Zengrui
Li, Zhaoqing
Lin, Long
Povey, Daniel
Audio and Speech Processing
Machine Learning
Sound
Connectionist Temporal Classification (CTC) is a widely used method for automatic speech recognition (ASR), renowned for its simplicity and computational efficiency. However, it often falls short in recognition performance. In this work, we propose the Consistency-Regularized CTC (CR-CTC), which enforces consistency between two CTC distributions obtained from different augmented views of the input speech mel-spectrogram. We provide in-depth insights into its essential behaviors from three perspectives: 1) it conducts self-distillation between random pairs of sub-models that process different augmented views; 2) it learns contextual representation through masked prediction for positions within time-masked regions, especially when we increase the amount of time masking; 3) it suppresses the extremely peaky CTC distributions, thereby reducing overfitting and improving the generalization ability. Extensive experiments on LibriSpeech, Aishell-1, and GigaSpeech datasets demonstrate the effectiveness of our CR-CTC. It significantly improves the CTC performance, achieving state-of-the-art results comparable to those attained by transducer or systems combining CTC and attention-based encoder-decoder (CTC/AED). We release our code at https://github.com/k2-fsa/icefall.
title CR-CTC: Consistency regularization on CTC for improved speech recognition
topic Audio and Speech Processing
Machine Learning
Sound
url https://arxiv.org/abs/2410.05101