Pengenalan Pola Fonem Vokal menggunakan Short Time Fourier Transform (STFT) dan Fitur Mel Frequency Cepstral Coefficient (MFCC)

Adriansyah, Ahmad rio
Prasetyo, Kurniawan Dwi
Atmam Al Faruqi, Hamdan Ainul

Jurnal Teknologi Terpadu
Vol 7 No 1: Juli, 2021

Phonemes are the building blocks of every oral language. Every utterance is composed of one or more phonemes. To improve the accuracy of acoustic models, the researchers attempted to identify the pattern of vowel phonemes in bahasa Indonesia using STFT and MFCC features. This paper analyzes 398 audio files gathered from 51 participants and explores the difference of phonemes a, i, u, e,o. Using SVM and Neural Network, the features are classified and tested. The result gave 93.8% accuracy using SVM with radial based kernel