Abstract 

 

The technology of Automatic Speech Recognition (ASR) can be embedded in hand-held devices or reached remotely through the use of server-based speech recognition approaches. We design and implement a ubiquitous speech recognition application for the Arabic language. We select a network-based model for the design and the deployment after a comparison made between speech recognition techniques. We analyze the performance of Arabic speakers through the Arabic speech corpus of isolated words. The experimental results are encouraging to improve the accuracy and efficiency of the proposed system despite the environmental challenges and degradation due to speech coding, the channel itself and the noises affecting speech recognition performance.

 

 

 Télécharger l'article : A Ubiquitous Application for Arabic Speech Recognition