Arabic vowels recognition based on wavelet average framing linear prediction coding and neural network

Abstract

In this work, an average framing linear prediction coding (AFLPC) technique for speaker-independent Arabic vowels recognition system was proposed. Usually, linear prediction coding (LPC) has been applied in many speech recognition applications, however, the combination of modified LPC termed AFLPC with wavelet transform (WT) is proposed in this study for vowel recognition. The investigation procedure was based on feature extraction and classification. In the stage of feature extraction, the distinguished resonance of vocal tract of Arabic vowel characteristics was extracted using the AFLPC technique. LPC order of 30 was found to be the best according to the system performance. In the phase of classification, probabilistic neural network (PNN) was applied because of its rapid response and ease in implementation. In practical investigation, performances of different wavelet transforms in conjunction with AFLPC were compared with one another. In addition, the capability analysis on the proposed system was examined by comparing with other systems proposed in latest literature. Referring to our experimental results, the PNN classifier could achieve a better recognition rate with discrete wavelet transform and AFLPC as a feature extraction method termed (LPCDWTF).