This paper studies some unique sounds of Arabic language from a speech recognition problem point of view.These are the pharyngeal and emphatic sounds set.This research also addresses some points and the degree of difficulty in recognizing these sounds,which are found uniquely in Arabic,to determine whether or not they will complicate or ease speech recognition processing compared to Latin languages.An in-depth study of pertinent existing literature indicated that these unique sounds have not been incorporated in any speech recognition study.Specifically,the system discussed in this paper incorporates speech acquisition modules in an integrated Arabic Speech Recognition System.Arabic alphabets were adapted as the training and testing data bases.Linear Predictive Coding (LPC) feature extraction techniques were developed,and pattern recognition was used as the main technique in the speech recognition algorithm.
展开▼
