India has diverse regional languages, due to which accent detection has become a crucial part of Automatic Speech Recognition (ASR) field. In this research, machine learning technique were used to detect and classify accent of Indian English language using Svarah dataset, created by AI4Bharat. Audio signals have some key acoustic features based on which we can identify an accent. For our research Mel-frequency cepstral coefficients (MFCCs) is used to extract these key features from an audio signal. These features are then passed into a machine learning model to predict the accent of a language. The machine learning model which is used in this research is Convolutional Neural Network (CNN) because it is effective in noise filtration and analyzing spectrograms to capture similar patterns, such as pitch and tone of speaking in an audio signal. After training the model for 50 epochs using the key extracted features from the MFCC, the model able to achieved an accuracy rate of 82.38%. From these results, we can conclude that MFCC works perfectly with CNN to predict the specific accents of a particular language.