All-pole models have been successful in speech processing. It is known that predictor coefficients can be recursively computed from cepstrum samples, but this estimate is not correct enough. Moreover, cepstrum computation is rather complex. The actual availability of Hardware Fourier transformers have stirred renewed interest in the homomorphic processing. In this paper an algorithm to improve the predictor identification from the cepstrum is presented; the algorithm has been tested in many natural and synthetic speech signals and some typical results are reported. Finally, a vocoder application of the algorithm is suggested.
Identification of an all-pole model of speech by a least squares homomorphic algorithm
MUMOLO, ENZO
1984-01-01
Abstract
All-pole models have been successful in speech processing. It is known that predictor coefficients can be recursively computed from cepstrum samples, but this estimate is not correct enough. Moreover, cepstrum computation is rather complex. The actual availability of Hardware Fourier transformers have stirred renewed interest in the homomorphic processing. In this paper an algorithm to improve the predictor identification from the cepstrum is presented; the algorithm has been tested in many natural and synthetic speech signals and some typical results are reported. Finally, a vocoder application of the algorithm is suggested.File in questo prodotto:
Non ci sono file associati a questo prodotto.
Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.