Linear Prediction Coefficients and Linear Predication Cepstral Coefficients have been used as the main features for speech processing. Since the 1980s, it has been common practice in speech processing to use the acoustic features offered by extracting the Melfrequency cepstral coefficients (MFCCs).
These coefficients make up Mel-frequency cepstral, which is a representation of the short-term power spectrum of a sound. This is based on a linear discrete cosine transform of the log power spectrum on a nonlinear mel scale of frequency.
MFCC is based on the known variation of our ears’ perception and sensitivity with respect to frequencies.