Robust Detection of Phone Boundaries Using Model Selection Criteria With Few Observations

被引:17
作者
Almpanidis, George [1 ]
Kotti, Margarita [1 ]
Kotropoulos, Constantine [1 ]
机构
[1] Aristotle Univ Thessaloniki, Dept Informat, Thessaloniki 54124, Greece
来源
IEEE TRANSACTIONS ON AUDIO SPEECH AND LANGUAGE PROCESSING | 2009年 / 17卷 / 02期
关键词
Automatic phonetic segmentation; model selection; robust statistics; SEGMENTATION;
D O I
10.1109/TASL.2008.2009162
中图分类号
O42 [声学];
学科分类号
070206 ; 082403 ;
摘要
Automatic phone segmentation techniques based on model selection criteria are studied. We investigate the phone boundary detection efficiency of entropy- and Bayesian- based model selection criteria In continuous speech based on the DISTBIC hybrid segmentation algorithm. DISTBIC is a text-independent bottom-up approach that identifies sequential model changes by combining metric distances with statistical hypothesis testing. Using robust statistics and small sample corrections in the baseline DISTBIC algorithm, phone boundary detection accuracy is significantly improved, while false alarms are reduced. We also demonstrate further improvement in phonemic segmentation by taking into account how the model parameters are related in the probability density functions of the underlying hypotheses as well as in the model selection via the information complexity criterion and by employing M-estimators or the model parameters. The proposed DISTBIC variants are tested on the NTIMIT database and the achieved F-1 measure is 74.7% using a 20-ms tolerance in phonemic segmentation.
引用
收藏
页码:287 / 298
页数:12
相关论文
共 47 条
  • [1] Adell J., 2004, P 5 ISCA SPEECH SYNT, P139
  • [2] Robust speaker change detection
    Ajmera, J
    McCowan, L
    Bourlard, H
    [J]. IEEE SIGNAL PROCESSING LETTERS, 2004, 11 (08) : 649 - 651
  • [3] Akaike H., 1992, Breakthroughs in Statistics, P610, DOI [10.1007/978-1-4612-0919-538, DOI 10.1007/978-1-4612-0919-538, DOI 10.1007/978-1-4612-1694-0_15]
  • [4] Phonemic segmentation using the generalised Gamma distribution and small sample Bayesian information criterion
    Almpanidis, George
    Kotropoulos, Constantine
    [J]. SPEECH COMMUNICATION, 2008, 50 (01) : 38 - 55
  • [5] ANTAL M, 2004, STUD U BABES BOLYAI, V49, P55
  • [6] Barnette J.J., 1998, ANN M MIDS ED RES AS
  • [7] Bearse PM, 1997, J APPL ECONOMET, V12, P563, DOI 10.1002/(SICI)1099-1255(199709/10)12:5<563::AID-JAE453>3.0.CO
  • [8] 2-V
  • [9] BHANSALI RJ, 1977, BIOMETRIKA, V64, P547, DOI 10.1093/biomet/64.3.547