The main objective of any character recognition system is to achieve modification  or conversion of any form of text such as printed or scanned text images. Most of the greatest literature works in ancient India were written in Sanskrit. Hence it is necessary to preserve these highly prestigious documents by digitizing them. For digitization of these documents it is necessary that these characters get accurately recognized by the machine for which it need an accurate classifier. In this paper shows comparison between various classifiers used for letter classification which are designed using machine learning and deep Learning algorithms 
 
Keywords: Thresholding, Classifier, SVM, Deep learning, CNN, Discrete wavelet transform I