7th International Conference on Spoken Language Processing

September 16-20, 2002
Denver, Colorado, USA

On the Use of Gaussian Mixture Model for Speaker Variability Analysis

Tao Chen (1), Chao Huang (2), Eric Chang (2), Jingchun Wang (1)

(1) Tsinghua University, China; (2) Microsoft Research Asia, China

Analysis and modeling of speaker variability is important to help understand in-depth inter-speaker variances and to enhance current speech/speaker recognition system. In this paper we introduce adapted Gaussian mixture model (GMM) based speaker representation for the task. Two powerful multivariate statistical analysis methods, principal component analysis (PCA) and independent component analysis (ICA), are used to extract the sources of dominant speaker variability. In addition, analysis of variance (ANOVA) is adopted to evaluate the dominance of a factor in a certain principal/independent component. Further, the generalization ability of our method is investigated by experiments.

Full Paper

Bibliographic reference.  Chen, Tao / Huang, Chao / Chang, Eric / Wang, Jingchun (2002): "On the use of Gaussian mixture model for speaker variability analysis", In ICSLP-2002, 1249-1252.