Optimization of Gaussian Mixture Model Parameters for Speaker Identification

Abstract. Gaussian mixture model (GMM) [1] has been widely used for modeling speakers. In speaker identification, one major problem is how to generate a set of GMMs for identification purposes based upon the training data. Due to the hill-climbing characteristic of the maximum likelihood (ML) method, any arbitrary estimate of the initial model pa-rameters will usually lead to a sub-optimal model in practice. To resolve this problem, this paper proposes a hybrid training method based on the genetic algorithm (GA). It utilizes the global searching capability of the GA and combines the effectiveness of the ML method. 1 The Proposed Algorithm Spoken utterances convey both linguistic and speaker-specific information. In the GMM-based speaker identification, the distribution of feature vectors extracted from a speaker’s utterance is modeled by a weighted sum of mixture components and the speaker model that has the highest likelihood score for the utterance will be selected as the identification result. Before the identification task, the mode