Probabilistic Classification Based on Gaussian Copula for Speech Recognition: Application to Spoken Arabic Digits

Language modeling for an inflected language such as Arabic poses new challenges for automatic speech recognition and related topic due to its rich morphology. A new technique for automatic speech recognition is presented in this paper. This technique employs a full measure of statistical dependence among random variables that is known as copulas. A novel probabilistic classifier that combines finite Gaussian mixture modeling for marginal distribution function and Gaussian copula is developed. Using benchmark Arabic speech data base, the accuracy of the developed Gaussian copula with Gaussian Mixtures marginal distribution GCGMM is validated and compared with Gaussian copula with simple empirical marginal distribution GCEM. The result demonstrates the improvement and shows an excellent performance.