Transformation-based GMM with improved cluster algorithm for speaker identification

The embedded linear transformation is a popular technique which integrates both transformation and diagonal-covariance Gaussian mixture into a unified framework to improve the performance of speaker recognition. However, the mixture number of GMM must be given in model training. The cluster expectation-maximization (EM) algorithm is a well-known technique in which the mixture number is regarded as an estimated parameter. This paper presents a new model that integrates an improved cluster algorithm into the estimating process of GMM with the embedded transformation. In the approach, the transformation matrix, the mixture number and other traditional model parameters are simultaneously estimated according to a maximum likelihood criterion. The proposed method is demonstrated on a database of three data sessions for text independent speaker identification. The experiments show that this method outperforms the traditional GMM with cluster EM algorithm.