Comparative Study of Amazigh Speech Recognition Systems Based on Different Toolkits and Approaches

Abstract

The objective of this study is to evaluate and contrast the performance of different ASR approaches applied to the Amazigh language. Markovian modelling techniques, including Hidden Markov Models with Gaussian mixture distribution, Convolutional Neural Network, size of vocabulary, and lastly, the choice of decoder, whether Sphinx or HTK, by conducting a comprehensive analysis and comparison of these factors, this paper aims to provide valuable insights into the development of effective ASR systems for the Amazigh language. The findings will contribute to advancing the field of Amazigh ASR and aid in the selection of appropriate techniques and tools for future research and development efforts.