In this paper, a new database suitable for HMM-based automatic Filipino speech recognition is described for the purpose of training a domain-independent, large-vocabulary continuous speech recognition system. Although it is known that high-performance speech recognition systems depend on a superior speech database used in the training stage, due to the lack of such an appropriate database, previous reports on Filipino speech recognition had to contend with serious data sparsity issues. In this paper we alleviate such sparsity through appropriate data analysis that makes the evaluation results more reliable. The best system is identified through its low word-error rate to a cross-validation set containing almost three hours of unknown speech data. Language-dependent problems are discussed, and their impact on accuracy was analyzed. The approach is currently data driven, however it serves as a competent baseline model for succeeding future developments.
Federico ANG
University of the Philippines
Rowena Cristina GUEVARA
University of the Philippines
Yoshikazu MIYANAGA
Hokkaido University
Rhandley CAJOTE
University of the Philippines
Joel ILAO
University of the Philippines
Michael Gringo Angelo BAYONA
University of the Philippines
Ann Franchesca LAGUNA
University of the Philippines
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copy
Federico ANG, Rowena Cristina GUEVARA, Yoshikazu MIYANAGA, Rhandley CAJOTE, Joel ILAO, Michael Gringo Angelo BAYONA, Ann Franchesca LAGUNA, "Open Domain Continuous Filipino Speech Recognition: Challenges and Baseline Experiments" in IEICE TRANSACTIONS on Information,
vol. E97-D, no. 9, pp. 2443-2452, September 2014, doi: 10.1587/transinf.2013EDP7442.
Abstract: In this paper, a new database suitable for HMM-based automatic Filipino speech recognition is described for the purpose of training a domain-independent, large-vocabulary continuous speech recognition system. Although it is known that high-performance speech recognition systems depend on a superior speech database used in the training stage, due to the lack of such an appropriate database, previous reports on Filipino speech recognition had to contend with serious data sparsity issues. In this paper we alleviate such sparsity through appropriate data analysis that makes the evaluation results more reliable. The best system is identified through its low word-error rate to a cross-validation set containing almost three hours of unknown speech data. Language-dependent problems are discussed, and their impact on accuracy was analyzed. The approach is currently data driven, however it serves as a competent baseline model for succeeding future developments.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2013EDP7442/_p
Copy
@ARTICLE{e97-d_9_2443,
author={Federico ANG, Rowena Cristina GUEVARA, Yoshikazu MIYANAGA, Rhandley CAJOTE, Joel ILAO, Michael Gringo Angelo BAYONA, Ann Franchesca LAGUNA, },
journal={IEICE TRANSACTIONS on Information},
title={Open Domain Continuous Filipino Speech Recognition: Challenges and Baseline Experiments},
year={2014},
volume={E97-D},
number={9},
pages={2443-2452},
abstract={In this paper, a new database suitable for HMM-based automatic Filipino speech recognition is described for the purpose of training a domain-independent, large-vocabulary continuous speech recognition system. Although it is known that high-performance speech recognition systems depend on a superior speech database used in the training stage, due to the lack of such an appropriate database, previous reports on Filipino speech recognition had to contend with serious data sparsity issues. In this paper we alleviate such sparsity through appropriate data analysis that makes the evaluation results more reliable. The best system is identified through its low word-error rate to a cross-validation set containing almost three hours of unknown speech data. Language-dependent problems are discussed, and their impact on accuracy was analyzed. The approach is currently data driven, however it serves as a competent baseline model for succeeding future developments.},
keywords={},
doi={10.1587/transinf.2013EDP7442},
ISSN={1745-1361},
month={September},}
Copy
TY - JOUR
TI - Open Domain Continuous Filipino Speech Recognition: Challenges and Baseline Experiments
T2 - IEICE TRANSACTIONS on Information
SP - 2443
EP - 2452
AU - Federico ANG
AU - Rowena Cristina GUEVARA
AU - Yoshikazu MIYANAGA
AU - Rhandley CAJOTE
AU - Joel ILAO
AU - Michael Gringo Angelo BAYONA
AU - Ann Franchesca LAGUNA
PY - 2014
DO - 10.1587/transinf.2013EDP7442
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E97-D
IS - 9
JA - IEICE TRANSACTIONS on Information
Y1 - September 2014
AB - In this paper, a new database suitable for HMM-based automatic Filipino speech recognition is described for the purpose of training a domain-independent, large-vocabulary continuous speech recognition system. Although it is known that high-performance speech recognition systems depend on a superior speech database used in the training stage, due to the lack of such an appropriate database, previous reports on Filipino speech recognition had to contend with serious data sparsity issues. In this paper we alleviate such sparsity through appropriate data analysis that makes the evaluation results more reliable. The best system is identified through its low word-error rate to a cross-validation set containing almost three hours of unknown speech data. Language-dependent problems are discussed, and their impact on accuracy was analyzed. The approach is currently data driven, however it serves as a competent baseline model for succeeding future developments.
ER -