This paper describes the development of an optimal sigmoidal rate-level function that is a part of many models of the peripheral auditory system. The optimization makes use of a set of criteria defined on the basis of physical attributes of the input sound that are inspired by physiological evidence. These criteria attempt to discriminate between a degraded speech signal and noise to preserve the maximum information in the linear region of the sigmoidal curve, and to minimize the effects of distortion in the saturating regions. The performance of the proposed approach is validated by text-independent speaker-verification experiments with signals corrupted by additive noise at different SNRs. Experimental results suggest that the approach presented in combination with CVN can lead to relative reductions in EER as great as 30% when compared with the use of baseline MFCC coefficients for some SNRs.
Bibliographic reference. Poblete, Víctor / Yoma, Néstor Becerra / Stern, Richard M. (2013): "Optimization of sigmoidal rate-level function based on acoustic features", In INTERSPEECH-2013, 896-900.