Source Feature Based Gender Identification System Using GMM

In this paper, through different experimental studies it is demonstrated that the excitation component of speech can be exploited for text independent gender identification system. Linear prediction (LP) residual is used as a representation of excitation information in speech. The speakerspecific information in the excitation of voiced speech is captured using Gaussian Mixture Model (GMM). The decrease in the error during training and recognizing correct gender during testing demonstrates that the excitation component of speech contains speaker-specific information and is indeed being captured by GMM. It is demonstrated that the proposed gender identification system using excitation information requires significantly less amount of data both during training as well as in testing, comparedto the other gender identification systems. A gender identification study using source feature for different Mixtures Components, train and test duration has been exploited. We demonstrate the gender identification studies on TIMIT database.