Accession Number : ADA261523


Title :   Supervised and Unsupervised Feature Extraction from a Cochlear Model for Speech Recognition


Descriptive Note : Technical rept.


Corporate Author : BROWN UNIV PROVIDENCE RI INST FOR BRAIN AND NEURAL SYSTEMS


Personal Author(s) : Intrator, N ; Tajchman, G


Full Text : https://apps.dtic.mil/dtic/tr/fulltext/u2/a261523.pdf


Report Date : 23 Dec 1992


Pagination or Media Count : 11


Abstract : We explore the application of a novel classification method that combines supervised and unsupervised training, and compare its performance to various more classical methods. We first construct a detailed high dimensional representation of the speech signal using Lyon's cochlear model and then optimally reproduce its dimensionality. The resulting low dimensional projection retains the information needed for robust speech recognition.


Descriptors :   *MODELS , *SPEECH RECOGNITION , NEURAL NETS , TRAINING , PERFORMANCE(HUMAN) , SPEECH , MOTOR NEURONS , LEARNING , EXTRACTION , SIGNALS , CLASSIFICATION , RECOGNITION


Subject Categories : Anatomy and Physiology
      Test Facilities, Equipment and Methods
      Voice Communications


Distribution Statement : APPROVED FOR PUBLIC RELEASE