Dan Reyes

Thesis Research: "AUDIO DEEPFAKE DETECTION THROUGH GLOTTAL FLOW ANALYSIS AND SPECTRAL FEATURE CLASSIFICATION". We developed a machine learning framework for audio deepfake detection using glottal flow and spectral feature analysis. Extracted 32 speech features (MFCC, LPC, Jitter, Shimmer, HNR, NAQ, and QOQ) using Parselmouth and Praat, and evaluated Support Vector Machine, Random Forest, and KNN classifiers. Achieved 96% file-level accuracy and 0.99 AUC in distinguishing real and AI-generated speech.