Text Independent Speaker Identification with Finite Multivariate Generalized Gaussian Mixture Model with Distant Microphone Speech

An effective and efficient speaker Identification (SI) system requires a robust feature extraction module followed by a speaker modeling scheme for generalized representation of these features. In recent, years Speaker Identification has seen significant advancement, but improvements have tended to be bench marked on the near field speech, ignoring the more realistic setting of far field instrumented speaker. A novel speaker model is developed by using Finite Multivariate Generalized Gaussian Mixture Model, Minimum Variance Distortion less Response Cepstral coefficients as feature Vectors. The performance of the developed model is studied through experimental evaluation with 45 speaker‟s data base and identification accuracy.