This paper describes techniques for modification and incorporation of excitation source features to convert neutral speech to emotional speech. Sad and anger are considered in this work. The features that have been used are epoch strength, epoch sharpness and pitch contour. A new method of modifying and incorporating epoch strength and sharpness is proposed. New pitch contours corresponding to the target emotions are obtained using two methods: Gaussian normalization and polynomial regression. The performance of these two methods is compared perceptually. The derived pitch contour is incorporated using time domain pitch synchronous overlap and add (PSOLA) method. Furthermore, different modification factors for epoch strengths and sharpness are used, and the resulting output is observed. The safe limits of these factors that produce no distortion in the output are also found out by experimenting, and factors outside these ranges are also studied for their effect on the resultant speech. Though the output does not completely resemble the target emotion, the potential of these methods to change the style of the speech is demonstrated by these experiments. These methods might turn out to be promising, if they are used along with vocal tract modification.