This document summarizes a research paper on speaker recognition using vocal tract features. It discusses extracting pitch using FFT as a vocal source feature and Mel Frequency Cepstral Coefficients (MFCCs) as vocal tract features. These features are used as inputs to a Support Vector Machine (SVM) classifier for binary speaker classification. Experimental results show the SVM achieves up to 94.42% accuracy in classifying speakers based on their MFCC features, and accuracy increases with higher signal-to-noise ratios when noise is added to speech signals. The paper concludes vocal source and tract features can be used together with SVM for text-independent speaker recognition.