This paper presents a parallel associative memory (PAM) method for phone estimation in speech recognition applications. The present method is the core of the SPEAR speech analysis and recognition tool developed at the University of Genova, Italy. The PAM architecture is explained in detail and compared in cost/performance tradeoff with other State of the Art systems. Representative experiments are reported to show the PAM’s operation under clean and noisy environments. The current tests show accurate phone recognition levels over a S/N rate >= 15 dB and good performance when compared to other phone estimators.
A Parallel Associative Memory Architecture for Phone Estimation
CURATELLI, FRANCESCO
1999-01-01
Abstract
This paper presents a parallel associative memory (PAM) method for phone estimation in speech recognition applications. The present method is the core of the SPEAR speech analysis and recognition tool developed at the University of Genova, Italy. The PAM architecture is explained in detail and compared in cost/performance tradeoff with other State of the Art systems. Representative experiments are reported to show the PAM’s operation under clean and noisy environments. The current tests show accurate phone recognition levels over a S/N rate >= 15 dB and good performance when compared to other phone estimators.File in questo prodotto:
Non ci sono file associati a questo prodotto.
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.