Metodo para entrenar un modelo acustico, que comprende realizar una alineacion forzada de datos de entrenamiento de voz, procesar los datos y obtener matrices estimadas de dispersion, sesgar las matrices de dispersion de clase intermedia e interna, diagonalizar las matrices de clase intermedia e interna y estimar vectores propios, obtener nuevas funciones discriminativas usando los vectores, entrenar un nuevo modelo acustico, guardar dicho modelo; metodo y sistema para formar modelo acustico

摘要

A system and method are presented for selectively biased linear discriminant analysis in automatic speech recognition systems. Linear Discriminant Analysis (LDA) may be used to improve the discrimination between the hidden Markov model (HMM) tied-states in the acoustic feature space. The between-class and within-class covariance matrices may be biased based on the observed recognition errors of the tied-states, such as shared HMM states of the context dependent tri-phone acoustic model. The recognition errors may be obtained from a trained maximum-likelihood acoustic model utilizing the tied-states which may then be used as classes in the analysis.