End-to-End-Based Tibetan Multitask Speech Recognition
To date, speech recognition technology for majority languages has been applied in wireless communication devices successfully. However, as a minority language, Tibetan has very limited resources for conventional automatic speech recognition. It lacks of enough data, sub-word units, lexicons, and wor...
Main Authors: | , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
IEEE
2019-01-01
|
Series: | IEEE Access |
Subjects: | |
Online Access: | https://ieeexplore.ieee.org/document/8894432/ |
id |
doaj-90fe08bc87f745689b50ebe605a9421b |
---|---|
record_format |
Article |
spelling |
doaj-90fe08bc87f745689b50ebe605a9421b2021-03-30T00:53:48ZengIEEEIEEE Access2169-35362019-01-01716251916252910.1109/ACCESS.2019.29524068894432End-to-End-Based Tibetan Multitask Speech RecognitionYue Zhao0https://orcid.org/0000-0002-4007-7016Jianjian Yue1Xiaona Xu2Licheng Wu3Xiali Li4School of Information and Engineering, Minzu University of China, Beijing, ChinaSchool of Information and Engineering, Minzu University of China, Beijing, ChinaSchool of Information and Engineering, Minzu University of China, Beijing, ChinaSchool of Information and Engineering, Minzu University of China, Beijing, ChinaSchool of Information and Engineering, Minzu University of China, Beijing, ChinaTo date, speech recognition technology for majority languages has been applied in wireless communication devices successfully. However, as a minority language, Tibetan has very limited resources for conventional automatic speech recognition. It lacks of enough data, sub-word units, lexicons, and word inventories for some dialects. In this paper, we present a multitask end-to-end model to perform simultaneous Tibetan speech content recognition, dialect identification, and speaker recognition. This model avoids processing the pronunciation dictionary and word segmentation for new dialects while allowing for training three tasks in a single model. We build the multitask recognition framework based on WaveNet-CTC. The dialect information and speaker ID are used in the output for training. The experimental results show that our method has better performance compared with a task-specific model.https://ieeexplore.ieee.org/document/8894432/End-to-end modelmultitask speech recognitionTibetan languagewavenet model |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Yue Zhao Jianjian Yue Xiaona Xu Licheng Wu Xiali Li |
spellingShingle |
Yue Zhao Jianjian Yue Xiaona Xu Licheng Wu Xiali Li End-to-End-Based Tibetan Multitask Speech Recognition IEEE Access End-to-end model multitask speech recognition Tibetan language wavenet model |
author_facet |
Yue Zhao Jianjian Yue Xiaona Xu Licheng Wu Xiali Li |
author_sort |
Yue Zhao |
title |
End-to-End-Based Tibetan Multitask Speech Recognition |
title_short |
End-to-End-Based Tibetan Multitask Speech Recognition |
title_full |
End-to-End-Based Tibetan Multitask Speech Recognition |
title_fullStr |
End-to-End-Based Tibetan Multitask Speech Recognition |
title_full_unstemmed |
End-to-End-Based Tibetan Multitask Speech Recognition |
title_sort |
end-to-end-based tibetan multitask speech recognition |
publisher |
IEEE |
series |
IEEE Access |
issn |
2169-3536 |
publishDate |
2019-01-01 |
description |
To date, speech recognition technology for majority languages has been applied in wireless communication devices successfully. However, as a minority language, Tibetan has very limited resources for conventional automatic speech recognition. It lacks of enough data, sub-word units, lexicons, and word inventories for some dialects. In this paper, we present a multitask end-to-end model to perform simultaneous Tibetan speech content recognition, dialect identification, and speaker recognition. This model avoids processing the pronunciation dictionary and word segmentation for new dialects while allowing for training three tasks in a single model. We build the multitask recognition framework based on WaveNet-CTC. The dialect information and speaker ID are used in the output for training. The experimental results show that our method has better performance compared with a task-specific model. |
topic |
End-to-end model multitask speech recognition Tibetan language wavenet model |
url |
https://ieeexplore.ieee.org/document/8894432/ |
work_keys_str_mv |
AT yuezhao endtoendbasedtibetanmultitaskspeechrecognition AT jianjianyue endtoendbasedtibetanmultitaskspeechrecognition AT xiaonaxu endtoendbasedtibetanmultitaskspeechrecognition AT lichengwu endtoendbasedtibetanmultitaskspeechrecognition AT xialili endtoendbasedtibetanmultitaskspeechrecognition |
_version_ |
1724187722660184064 |