Phonics Recognition Using Machine Learning with MFCC Extraction

Phonics is an educational method designed to help children decode written letters and spoken sounds, ultimately aiding in reading proficiency. Currently, there is no research on Phonics alphabet recognition. Thus, this research explores the application of Machine Learning techniques for recognizing Phonics alphabet spelling, utilizing Mel-Frequency Cepstral Coefficients (MFCC) for feature extraction, known for its effectiveness in speech recognition. The research employs classifiers such as Support Vector Machine (SVM), K-Nearest Neighbor (KNN), and Random Forest, incorporating data augmentation techniques like random and constant noise addition and time shifting, with MFCC extraction performed both manually and via the Librosa library. The results demonstrate that the SVM model, when combined with constant data augmentation and MFCC extraction using Librosa, achieved the highest accuracy of 95.57%, highlighting its superior performance in Phonics alphabet spelling recognition.