Comparative Study of ANN and HMM to Arabic Digits Recognition Systems

Arabic language is a Semitic language that has manydifferences when compared to Latin languages such as English. Oneof these differences is how to pronounce the ten digits, zero throughnine. All Arabic digits are polysyllabic (except digit zero which is amonosyllabic) words and most of them contain Arabic uniquephonemes, namely, pharyngeal and emphatic subset. In a previouspaper the researcher designed an Artificial Neural Networks (ANN)based Arabic digits recognition system. In this paper we continued theresearch by designing Hidden Markov Model (HMM) based systemthat was designed and tested with automatic Arabic digits recognition.The old system was isolated whole word speech recognizer, but thecurrent one was an isolated word phoneme based recognizer. Bothsystems were implemented both as a multi-speaker (i.e., the same setof speakers were used in both the training and testing phases) modeand speaker-independent (i.e., speakers used for training are differentfrom those used for testing) mode. The main aim of this paper was tocompare, analyze, and discuss the outcomes of these two recognitionsystems. The ANN based recognition system achieved 99.5% correctdigit recognition in the case of multi-speaker mode, and 94.5% in thecase of speaker-independent mode. On the other hand, the HMMbased recognition system achieved 98.1% correct digit recognition inthe case of multi-speaker mode, and 94.8% in the case of speakerindependentmode.