In this paper a method for Voiced/Unvoiced classification of segments of a speech signal is presented. For this purpose, on each segment four different measurements are made. On the basis of such measurements the a posteriori probabilities of the two classes are determined and the decision is made by means of the maximum a posteriori probability criterion. The a priori probabilities of the classes are updated for each segment. In this context, it is assumed that the sequence of the classes constitutes a 1st order time-omogeneous Markov chain. The experimental result of the method, included in the paper, have been satisfactory.
A Bayesian approach for voiced/unvoiced classification of segments of a speech signal / Bruno, Giordano; M. D., Di Benedetto; DI BENEDETTO, Maria Gabriella; Gilio, Angelo. - STAMPA. - I:(1981), pp. 181-184. (Intervento presentato al convegno The Fourth F.A.S.E. Symposium on Acoustic and Speech tenutosi a Venezia nel 21-24 Aprile 1981).
A Bayesian approach for voiced/unvoiced classification of segments of a speech signal
BRUNO, Giordano;DI BENEDETTO, Maria Gabriella;GILIO, ANGELO
1981
Abstract
In this paper a method for Voiced/Unvoiced classification of segments of a speech signal is presented. For this purpose, on each segment four different measurements are made. On the basis of such measurements the a posteriori probabilities of the two classes are determined and the decision is made by means of the maximum a posteriori probability criterion. The a priori probabilities of the classes are updated for each segment. In this context, it is assumed that the sequence of the classes constitutes a 1st order time-omogeneous Markov chain. The experimental result of the method, included in the paper, have been satisfactory.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
