PHONEME (VOWEL SOUND) RECOGNITION USING MACHINE LEARNING

Abstract

Currently, Automatic Speech Recognition (ASR) systems have become increasingly relevant due to advances in artificial intelligence and machine learning. The relevance of this study lies in the insufficient development of phoneme recognition systems for the Kazakh language, particularly for vowel sounds, and the limited availability of digital linguistic resources. The aim of the research is to develop a machine learning model capable of accurately recognizing Kazakh vowel sounds. The study involved preprocessing of speech signals, feature extraction using MFCC, and comparison of the Random Forest, SVM, and ANN algorithms. As a result, the ANN model achieved the highest accuracy. The findings can enhance the quality of Kazakh speech recognition systems and be applied in voice biometrics and sound interface technologies.