Accession Number : ADA300991
Title : Graphical Analysis of Hidden Markov Model Speech Recognition Experiments.
Descriptive Note : Technical rept.,
Corporate Author : MASSACHUSETTS INST OF TECH LEXINGTON LINCOLN LAB
Personal Author(s) : Seward, D. C. ; Zissman, M. A.
PDF Url : ADA300991
Report Date : 25 OCT 1995
Pagination or Media Count : 53
Abstract : Hidden Markov models are powerful tools for acoustic modeling III speech recognition systems. However, detailed analysis of their performance in specific experiments can be difficult. Two tools were developed and implemented for the purpose of analyzing hidden Markov model experiments: an interactive Viterbi backtrace viewer and a multidimensional scaling display. These tools were built using the HMM Toolkit. Use of the Viterbi backtrace tool provided insight that eventually led to improved recognition performance. (AN)
Descriptors : *MATHEMATICAL MODELS, *SPEECH RECOGNITION, *PATTERN RECOGNITION, DATA BASES, ALGORITHMS, SOFTWARE ENGINEERING, NEURAL NETS, DATA MANAGEMENT, LEARNING MACHINES, INPUT OUTPUT PROCESSING, PROBABILITY DENSITY FUNCTIONS, SCALING FACTOR, WORDS(LANGUAGE), SPEECH ANALYSIS, ACOUSTIC DATA, MARKOV PROCESSES, WORD RECOGNITION, PHONEMES.
Subject Categories : Computer Programming and Software
Distribution Statement : APPROVED FOR PUBLIC RELEASE