Evolutionary learning of HMM with Gaussian mixture densities for Automatic speech recognition
Abdelmadjid Benmachiche, Amina Makhlouf, Tahar Bouhadada · 2019
Automatic speech recognition (ASR) is a very important field that can be used in many applications such as facilitate a physically handicapped person to command and control a machine, telecommunication systems, investigations, voice email and management. ASR systems are becoming more and more important today. In speech recognition, the training method plays a very important role. When a good practice model is obtained for speech mode, it means that it increases the speed of recognition significantly. In this paper, we explore the possibility of using Genetic Algorithms (GAs) for optimizing the parameter learning of hidden Markov models (HMM) in automatic speech recognition systems. This approach has been tested on samples from different users with different environments (with/without adding noise) and achieves a high degree of accuracy during recognition.