Control Interface of an Automatic Continuous Speech Recognition System in Standard Arabic Language

Abstract

Human-machine dialogue is still too difficult to set up, but remains an important issue for the Arabic language, which remains a language poorly endowed with communication systems. To this end, we have oriented our work around a control interface for an Automatic Continuous Speech Recognition System (ACSRS), which we have adapted to the Standard Arabic language. For this, a construction of an Automatic Continuous Speech Recognition System was first undertaken and a link was established with a graphic interface based on HMM (Hidden Models of Markov) and the toolbox HTK (Hidden Markov Model Toolkit). The realization of the proposed interface will avoid the dependency constraint, common in most of the ACSRS (s), and make the system independent of the speakers. On the one hand, the transcription of the sentences recognized by this system will be executed in real time; and on the other hand, the display will be on a screen independent of the system.