Automatic learning: An approach to the adaptation of a speech recognition system to one or several speakers

Abstract As part of a system for the automatic recognition of isolated words in a large vocabulary on the basis of an analytical approach, we considered the automatic speaker-adaptation of the system. This was carried out by means of an automatic learning procedure of the speakers' reference patterns, and by automatically adjusting the parameters of the system. This learning relies on a time alignment algorithm using acoustic-phonetic features which are little speakerdependent. The learning session was successfully tested on 18 speakers out of 20 (10 women and 10 men) and the reference patterns thus obtained yielded good results during the recognition phase. We have now undertaken an analysis of the vowels uttered by 15 speakers based upon descriptive statistics and statistical interpretation in order to design procedures of normalization and of automatic generation of a speaker's vowel reference patterns.

[1]  Michael Wagner.  Automatic labelling of continuous speech with a given phonetic transcription using dynamic programming algorithms , 1981, ICASSP.

[2]  L. Rabiner,et al.  A simplified, robust training procedure for speaker trained, isolated word recognition systems , 1980 .

[3]  John S. Bridle,et al.  Automatic labelling of speech using synthesis-by-rule and non-linear time-alignment , 1983, Speech Commun..

[4]  Matthew Lennig.  Automatic alignment of natural speech with a corresponding transcription , 1983, Speech Commun..

[5]  Sadaoki Furui,et al.  A training procedure for isolated word recognition systems , 1980 .

[6]  B. Lowerre,et al.  Dynamic speaker adaptation in the Harpy speech recognition system , 1977 .

[7]  Hartmut Traunmüller,et al.  Articulatory and perceptual factors controlling the age- and sex-conditioned variability in formant frequencies of vowels , 1984, Speech Commun..

[8]  N. Carbonell,et al.  An expert system for the automatic reading of French spectrograms , 1984, ICASSP.

[9]  Maxine Eskénazi,et al.  Cadrage automatique pour la constitution de dictionnaires d'entites phonetiques , 1983, Speech Commun..

[10]  Mario Rossi,et al.  Indices acoustiques multilocuteurs et independants du contexte pour la reconnaissance automatique de la parole , 1983, Speech Commun..

[11]  Mats Blomberg,et al.  Automatic time alignment of speech with a phonetic transcription , 1985 .

[12]  André Rigault,et al.  Sources of Inter- and Intra-Speaker Variability in the Acoustic Properties of Speech Sounds , 1972 .

[13]  G. E. Peterson,et al.  Control Methods Used in a Study of the Vowels , 1951 .

[14]  H. Wakita.  Normalization of vowels by vocal-tract length and its application to vowel identification , 1977 .

[15]  Lawrence R. Rabiner,et al.  Considerations in applying clustering techniques to speaker independent word recognition , 1979, ICASSP.

[16]  C. Weinstein,et al.  A system for acoustic-phonetic analysis of continuous speech , 1975 .

[17]  T. M. Nearey.  Phonetic feature systems for vowels , 1978 .

[18]  Hong C. Leung,et al.  Automatic alignment of phonetic transcriptions with continuous speech , 1984 .

[19]  Kiyohiro Shikano,et al.  Isolated word recognition using phoneme-like templates , 1983, ICASSP.

[20]  Jean-François Mari,et al.  Some experiments in automatic recognition of a thousand word vocabulary , 1984, ICASSP.