Extraction of specific data from a sound sample by removing additional distortion

Correct identity recognition based on a voice sample must deal with many problems such as too big or small distance from the microphone, noise or abnormal voice. Hoarseness, coughing or even stuttering can also be encountered as disturbance of the voice. Research on new aspects of intelligent processing for voice brings possibilities to use intelligent methods to increase efficiency in processing and quality of record. In this paper, a spectrogram analysis for the detection of specific data and remove these distortions in the sample is presented. The proposed solution has been tested and discussed for real use in identity verification systems.