“…The most commonly used spectral features for emotion recognition are Mel-Frequency Cepstral Coefficients (MFCC) [35, 16, 15, 22, 29, 19, 11, 37, 8, 30, 21, 27, 14, 12, 32, 33, 36, 38, 39]. As in automatic speech recognition, MFCC are extracted using a 25 ms Hamming window at intervals of 10 ms and cover frequency range from 300 Hz to the Nyquist frequency.…”