Provided by:International Journal of Advanced Research in Electrical, Electronics and Instrumentation Engineering (IJAREEIE)
This paper adopted Hidden Markov Model (HMM) to recognize Arabic isolated words. The database contains eleven Arabic isolated words. The authors have repeated each of them twenty five times by mono-locutor. Feature extraction using Bionic Wavelet Transform (BWT) and Mel Frequency Cepstral Coefficient (MFCC) are carried over the speech frame of the input speech. This is followed by Vector Quantization (VQ) and Hidden Markov Modeling. In this paper, they describe the use of Genetic Algorithm (GA) to make a global search of optimal HMM parameters.