Cover Image

A two stage adaptation towards automatic speech recognition system for children

Recently, Automatic Speech Recognition (ASR) systems were used to assist children in language acquisition as it has the ability to detect human speech signal. Despite the benefits offered by the ASR system for children, there is a lack of ASR systems for Malay-speaking children. One of the contr...

Full description

Saved in:
Bibliographic Details
Main Authors: Mumtaz Begum, M., Siti Salwah, S., Feizal Dani, R.
Format: Conference or Workshop Item
Language:English
Published: 2016
Subjects:
Q Science (General)
QA75 Electronic computers. Computer science
Online Access:http://eprints.um.edu.my/15793/1/0001.pdf
http://eprints.um.edu.my/15793/
Tags: Add Tag
No Tags, Be the first to tag this record!
id my.um.eprints.15793
record_format eprints
spelling my.um.eprints.157932016-04-20T02:25:15Z http://eprints.um.edu.my/15793/ A two stage adaptation towards automatic speech recognition system for children Mumtaz Begum, M. Siti Salwah, S. Feizal Dani, R. Q Science (General) QA75 Electronic computers. Computer science Recently, Automatic Speech Recognition (ASR) systems were used to assist children in language acquisition as it has the ability to detect human speech signal. Despite the benefits offered by the ASR system for children, there is a lack of ASR systems for Malay-speaking children. One of the contributing factors for this is the lack of continuous speech database for Malay-speaking children. Though cross-lingual adaptation is a common solution for developing ASR systems for under-resourced language, it is not viable for children as there are very limited children's speech databases as a source model. In this research, we are proposing a two-stage adaptation using a very limited database of Malay-speaking children for the development of the children ASR system. The two stage adaptation comprises the cross-lingual adaptation (first stage) and cross-age adaptation. For the first stage, a well-known speech database that is phonetically rich and balanced, is adapted to the medium-sized Malay adults using supervised MLLR. The second stage adaptation uses the speech acoustic model generated from the first adaptation, and the target database is a small-sized database of Malay-speaking children. We have measured the performance of the proposed technique using word error rate, and then compare them with the conventional benchmark adaptation. The two stage adaptation proposed in this research has better recognition accuracy as compared to the benchmark adaptation in recognizing children's speech. 2016 Conference or Workshop Item PeerReviewed application/pdf en http://eprints.um.edu.my/15793/1/0001.pdf Mumtaz Begum, M. and Siti Salwah, S. and Feizal Dani, R. (2016) A two stage adaptation towards automatic speech recognition system for children. In: International Conference on Text, Speech and Dialogue (ICTSD), 03 - 04 March 2016, Singapore.
institution Universiti Malaya
building UM Library
collection Institutional Repository
continent Asia
country Malaysia
content_provider Universiti Malaya
content_source UM Research Repository
url_provider http://eprints.um.edu.my/
language English
topic Q Science (General)
QA75 Electronic computers. Computer science
spellingShingle Q Science (General)
QA75 Electronic computers. Computer science
Mumtaz Begum, M.
Siti Salwah, S.
Feizal Dani, R.
A two stage adaptation towards automatic speech recognition system for children
description Recently, Automatic Speech Recognition (ASR) systems were used to assist children in language acquisition as it has the ability to detect human speech signal. Despite the benefits offered by the ASR system for children, there is a lack of ASR systems for Malay-speaking children. One of the contributing factors for this is the lack of continuous speech database for Malay-speaking children. Though cross-lingual adaptation is a common solution for developing ASR systems for under-resourced language, it is not viable for children as there are very limited children's speech databases as a source model. In this research, we are proposing a two-stage adaptation using a very limited database of Malay-speaking children for the development of the children ASR system. The two stage adaptation comprises the cross-lingual adaptation (first stage) and cross-age adaptation. For the first stage, a well-known speech database that is phonetically rich and balanced, is adapted to the medium-sized Malay adults using supervised MLLR. The second stage adaptation uses the speech acoustic model generated from the first adaptation, and the target database is a small-sized database of Malay-speaking children. We have measured the performance of the proposed technique using word error rate, and then compare them with the conventional benchmark adaptation. The two stage adaptation proposed in this research has better recognition accuracy as compared to the benchmark adaptation in recognizing children's speech.
format Conference or Workshop Item
author Mumtaz Begum, M.
Siti Salwah, S.
Feizal Dani, R.
author_facet Mumtaz Begum, M.
Siti Salwah, S.
Feizal Dani, R.
author_sort Mumtaz Begum, M.
title A two stage adaptation towards automatic speech recognition system for children
title_short A two stage adaptation towards automatic speech recognition system for children
title_full A two stage adaptation towards automatic speech recognition system for children
title_fullStr A two stage adaptation towards automatic speech recognition system for children
title_full_unstemmed A two stage adaptation towards automatic speech recognition system for children
title_sort two stage adaptation towards automatic speech recognition system for children
publishDate 2016
url http://eprints.um.edu.my/15793/1/0001.pdf
http://eprints.um.edu.my/15793/
_version_ 1643690137977094144
score 13.211869

Similar Items