Small vocabulary isolated-word automatic speech recognition for single-word commands in Arabic spoken

Research into automated speech recognition (ASR) for the Arabic language has been steadily increasing due to its potential for great growth. In this paper, we implemented Dynamic Time Warping (DTW) and Vector Quantization (VQ) techniques to apply to limited vocabulary speech recognition applications. Our goal was to build a small vocabulary, speaker independent isolated-word recognition system with a higher success rate for recognizing numerals between 0 and 9 in Palestinian-spoken Arabic. To do so, we enhanced the adopted methodology and improved its operator dependability. The algorithm obtained an accuracy rate of 99.6%. To achieve this, we will use the Mel Frequency Cepstral Coefficients (MFCC) algorithm to extract certain features from the speech, which will be used to reduce the dimensionality of the input voice. This algorithm will then be downloaded to a Digital Signal Processor (DSP) card, which will be responsible for recognizing the one-digit number and sending commands to the external world.