APSIPA Transactions on Signal and Information Processing > Vol 9 > Issue 1

End-to-end recognition of streaming Japanese speech using CTC and local attention

Jiahao Chen, Tokushima University, Japan, Ryota Nishimura, Tokushima University, Japan, Norihide Kitaoka, Toyohashi University of Technology, Japan, kitaoka@tut.jp
 
Suggested Citation
Jiahao Chen, Ryota Nishimura and Norihide Kitaoka (2020), "End-to-end recognition of streaming Japanese speech using CTC and local attention", APSIPA Transactions on Signal and Information Processing: Vol. 9: No. 1, e25. http://dx.doi.org/10.1017/ATSIP.2020.23

Publication Date: 23 Nov 2020
© 2020 Jiahao Chen, Ryota Nishimura and Norihide Kitaoka
 
Subjects
 
Keywords
CTCLocal attentionSpeech recognitionStreaming recognition
 

Share

Open Access

This is published under the terms of the Creative Commons Attribution licence.

Downloaded: 1026 times

In this article:
I. INTRODUCTION 
II. E2E SPEECH RECOGNITION 
III. DETAILS OF OUR APPROACH 
IV. EXPERIMENTAL SETUP 
V. RESULTS 
VI. CONCLUSIONS 

Abstract

Many end-to-end, large vocabulary, continuous speech recognition systems are now able to achieve better speech recognition performance than conventional systems. Most of these approaches are based on bidirectional networks and sequence-to-sequence modeling however, so automatic speech recognition (ASR) systems using such techniques need to wait for an entire segment of voice input to be entered before they can begin processing the data, resulting in a lengthy time-lag, which can be a serious drawback in some applications. An obvious solution to this problem is to develop a speech recognition algorithm capable of processing streaming data. Therefore, in this paper we explore the possibility of a streaming, online, ASR system for Japanese using a model based on unidirectional LSTMs trained using connectionist temporal classification (CTC) criteria, with local attention. Such an approach has not been well investigated for use with Japanese, as most Japanese-language ASR systems employ bidirectional networks. The best result for our proposed system during experimental evaluation was a character error rate of 9.87%.

DOI:10.1017/ATSIP.2020.23