Speech Denoinsing and Arabic Speaker Recognition System Using Subband Approach
Zied Sakka, Abdennaceur Kachouri, M. Samet · International Journal on Information Technology (IREIT) · 2016
This paper proposes an efficient speech recognition method for Arabic language. A Hidden Markov Models based speech recognition system was designed and tested with automatic Arabic word recognition. The system is an isolated whole word speech recognizer and it was implemented as both a wideband speech signal and a subbands spectral recognition modes. We particularly discuss the selection of the most critical subbands for the speaker recognition task and the choice of an optimal division of the frequency domain. An appropriate selection of the most critical subbands shows that very good performances are still obtained with only half of the frequency domain, the strategy of decision rests on the individual decisions of recognizers in each subband. This recognition system achieved a 89.5% correct word recognition in the wideband mode, and 95.25% in subbands mode. A comparison between the various variants of analysis will be made to observe their performances.