Title :
Automatic discovery of features underlying the perception of voicing
Author_Institution :
Dept. of Electron. & Comput. Sci., Southampton Univ., UK
Abstract :
Responses of both human and animal listeners to synthetic stop-consonant/vowel stimuli in which voice-onset time (VOT) is uniformly varied are known to be `categorical´ but an explanation of this phenomenon remains elusive. A `composite´ model consisting of a physiologically-realistic auditory model feeding its patterns of neural firing to an artificial neural network is described. When trained by (supervised) error back-propagation on the extreme, end-points of the VOT continuum, the composite model is capable of reproducing closely listeners´ behaviour in classical categorical-perception (CP) studies. However, whether the model also reproduces the so-called boundary-shift phenomenon-whereby the phoneme boundary moves with place of articulation-apparently depends upon the precise details of the auditory model and so, by implication, upon subtle aspects of peripheral auditory processing. A first attempt at unsupervised training has been unsuccessful: the likely reason for this is outlined. It is anticipated that future work comparing the model´s responses for unsupervised versus supervised training will help to elucidate the mechanisms of categorical perception
Keywords :
backpropagation; hearing; neural nets; physiological models; speech processing; unsupervised learning; animal listeners; artificial neural network; auditory model; boundary-shift phenomenon; categorical perception studies; composite model; error backpropagation; human listeners; neural firing patterns; peripheral auditory processing; phoneme boundary; physiologically-realistic auditory model; supervised training; synthetic stop-consonant/vowel stimuli; unsupervised training; voice-onset time; voicing; Animals; Artificial neural networks; Biological system modeling; Computer science; Heart; Humans; Intelligent systems; Intersymbol interference; Speech; Tail;
Conference_Titel :
Acoustics, Speech, and Signal Processing, 1995. ICASSP-95., 1995 International Conference on
Conference_Location :
Detroit, MI
Print_ISBN :
0-7803-2431-5
DOI :
10.1109/ICASSP.1995.479730