Coronary artery disease (CAD) is becoming the top cause of mortality worldwide. The major concern over CAD is its minimal preliminary symptoms arriving out of blocked coronary arteries. To enable early detection of underlying CAD heart sounds could be considered as potential indicators. Recent past studies give strong evidence for the same. This paper focuses on exploring the possibility of using spectral and Spectro temporal techniques to analyze heart sounds to detect coronary artery disease. Recording of heart sounds is done using electronic stethoscope 3M Littmann 3200. After verifying the fidelity of electronic stethoscope, spectral and Spectro temporal techniques are used to extract the features of heart sounds. The results from spectral analysis of heart sounds were utilized as a commencement point for Spectro-temporal feature analysis. Band wise spectral kurtosis is computed for after taking Stockwell transform for 10-400 Hz range. After experimentation, it was observed that band 3 (169-253 Hz) discriminated CAD and non-CAD heart sounds with linear separability. Minimum distance classifier has provided the classification accuracy of 95.64%. Therefore, preliminary diagnosis of Coronary artery Disease can be carried even in remote areas without costly equipment using this approach and provides major breakthrough for early diagnosis and prevention of mortalities for this dreaded ailment.