This paper introduces two significant contributions: one is a new feature based on histograms of MFCC (Mel-Frequency Cepstral Coefficients) extracted from the audio files that can be used in emotion classification from ...