Identifying and analysing the phonetics of low resources Nyishi language using artificial neural network technique

Abstract

Phonemic identification is an analysis and classification based on sounds. Phonetics can be categorized into three various types based on how it is articulated, perceived, and produced. The study introduces new Niyshi phonetic identification approaches based on artificial neural networks to improve the detectability of the system. This work has the goal of coming up with an effective technique that can be used to detect both non-nasalized and nasalized phonemes; by classifying phonemes into those belonging or not originating from the Nyishi dialect. The model that can be used to train, test, and validate can include a multilayer-based perceptron classifier. The proposed model identifies a complex old language with an accuracy as high as 87% through Multi-Layer Perceptron (MLP). To end, a confusion matrix analysis for ten trials involving an average of 89% accuracy for 20% trained dataset validation purposes confirms the efficiency and efficacy of the methodologies employed.