Paper Details

This paper describes the insight of the design & development of a Proposed Hindi Speech Recognizer basedon the continuous density hidden Markov model (CDHMM) . Here we have proposed a new recognizer which have been usedwith continuous density hidden Markov modeling to get a proposed CDHMM Hindi Speech Recognizer. Themultidimensional Mel frequency cepstral coefficients(MFCC) speech vectors are extracted from raw speech for every givenword that are used as a sequence of observation vectors, are uniformly segmented into 6 states . For each state GaussianMixture Model(GMM) parameters such as a Mean( jk) & Covariance ( jk) matrix & number of Mixtures(Cjk) arecalculated and simultaneously hidden Markov Model parameters such as   ( , , ) A Bare calculated to prepare a GMMHMM Model known as cdhmm , { , , }    Cjk jk jk  . Here ,Q=6, Mixture Components(K or M)=16, The covariancematrix used can be full or diagonal type matrix. Investigaions are done in this paper to find the optimal number of Gaussianmixture components that gives maximum accuracy in the context of Hindi speech recognition system. The results of theexperimentation have shown that Proposed CDHMM Speech Recognizer gives maximum performance when no. of GaussianMixture Model Components used is 16. This method is more powerful and efficient as compared to discrete SpeechRecognizer.

Satish Kumar, Prof. Jai Prakash, “DESIGN & DEVELOPMENT OF CONTINUOUS DENSITY HMM (CDHMM) ISOLATED HINDI SPEECH RECOGNIZER”, International Journal of Advance Engineering and Research Development (IJAERD), Vol. 2, Issue 5, pp. 1352-1357, May 2015.