Feature extraction method human factor cepstral coefficients in automatic speech recognition

Abstract

Using the Mel-frequency cepstral coefficients (MFCC), Human Factor cepstral coefficients (HFCC) and the modified technique of HFCC with gammachirp containing frequency domain noise and speech detection, these features are widely used for speech recognition in various applications. In speech recognition systems MFCC and HFCC are the two main techniques used. It will be shown in this paper that it presents some modifications to the original HFCC method. In our work the effectiveness of proposed changes to HFCC called Modified Human Factor cepstral coefficients (MHFCC) were tested and compared against the original HFCC features. The prosodic features such as jitter and shimmer are added to baseline spectral features. The above-mentioned techniques were tested with impulsive signals under various noisy conditions within AURORA databases.