Title :
Discriminative training of GMM for speaker identification
Author :
Álamo, C. Martín del ; Gil, F. J Caminero ; dela Torre Munilla, C. ; Gómez, L. Hernández
Author_Institution :
Telefonica Investigacion y Desarrollo, Madrid, Spain
Abstract :
We describe a novel discriminative training procedure for a Gaussian mixture model (GMM) speaker identification system. The proposal is based on the segmental generalized probabilistic descent (GPD) algorithm formulated to estimate the GMM parameters. Two major innovations over similar formulations of segmental GPD training are proposed. (1) A misclassification measure based on an individual representation of competing speakers, that explicitly allows to take into account different learning strategies for correctly or incorrectly classified speakers. (2) An empirical loss function to control the training procedure convergence, with a likelihood-based selection of correctly or incorrectly classified competing speakers. A comparison between the proposed method and the traditional GPD algorithm is also presented
Keywords :
Gaussian processes; parameter estimation; probability; speaker recognition; GMM; Gaussian mixture model; discriminative training; empirical loss function; learning strategies; likelihood-based selection; misclassification measure; parameter estimation; peaker identification system; segmental GPD training; segmental generalized probabilistic descent; speaker representation; training procedure convergence; Gas insulated transmission lines; Hidden Markov models; Iterative algorithms; Loss measurement; Maximum likelihood estimation; Parameter estimation; Proposals; Speech; Technological innovation; Training data;
Conference_Titel :
Acoustics, Speech, and Signal Processing, 1996. ICASSP-96. Conference Proceedings., 1996 IEEE International Conference on
Conference_Location :
Atlanta, GA
Print_ISBN :
0-7803-3192-3
DOI :
10.1109/ICASSP.1996.540297
