Accession Number : ADA311661

Title :   A Modulation Model for Characterizing Speech Signals.

Descriptive Note : Final technical rept. Sep 91-Apr 95,

Corporate Author : RUTGERS - THE STATE UNIV PISCATAWAY NJ

Personal Author(s) : Mammone, Richard J. ; Assaleh, Khaled ; Farrell, Kevin ; Ramachandran, Ravi ; Zilovic, Mihailo

PDF Url : ADA311661

Report Date : MAR 1996

Pagination or Media Count : 57

Abstract : The purpose of this effort was to research and develop new techniques for modeling speech segments. These models were used for Speaker Identification and for Keyword Spotting. The Format Mode Modulation Model and Pitch Mode Modulation Model were investigated. A novel feature, the Advanced Cepstral Weighting (ACW) Cepstrum, was developed and implemented to enhance performance. These algorithms were implemented on workstations, and also on an nCUBE supercomputer. In the future, these research algorithms can be used as part of a larger application.

Descriptors :   *SPEECH RECOGNITION, *CEPSTRUM TECHNIQUE, ALGORITHMS, IDENTIFICATION, WORK STATIONS, SUPERCOMPUTERS, MODULATION, INDEX TERMS.

Subject Categories : Voice Communications

Distribution Statement : APPROVED FOR PUBLIC RELEASE