Speaker recognition based on dynamic MFCC parameters

Abstract

The Mel-frequency cepstral coefficient is the most widely used feature in speech and speaker recognition. However, the traditional MFCC is very sensitive to noise interference, which tends to drastically degrade the performance of recognition systems because of the mismatches between training and testing. In this paper, we proposed a new speaker recognition algorithm based on the dynamic MFCC parameters. As the human auditory system can sensitively perceive the pitch changes in the speech, the algorithm, which combines the speaker information obtained by the MFCC with the pitch, can dynamically construct a set of Mel-filters according to the results of pitch detection. The Mel-filters are then used to extract the dynamic MFCC parameter, which represents the speaker's identity characteristics, and enhance accuracy of speaker recognition. The experimental results show that the method can perform well in a real environment and improve much on robustness in a noisy environment. The recognition rate in different signal-to-noise ratio conditions is obviously excelled to that of traditional MFCC with 5 to 6 percentage points higher on average.