Author

Dr. Engineer: Hassan Mohamed Ahmed

Abstract

The study of the independent text-independent method to determine the identity of the person by using its Voice Recognition voice, which is based on the extraction of features / features, which characterizes the linear autocorrelation function of the linear prediction behavior of the Cepstrum. The vector model of the person is built on a vector-based features. The Vector features a more reasonable blend of units that are known as the Maximally Plausible Mixture of Gaussians Gauss features. The identification process was carried out by voice in the form of a Maximum A Posteriori Probability model to be retrieved based on the audio input signal.
The proposed and proposed method presents the high and sufficient accuracy to determine the personality of the speaker by sound and in a separate text, compared with the results obtained at the global level in such systems

keywords

Voice tag, tone of voice, identification of person identity, verification of person's identity, sound characteristics, Spstrom audio signal, acoustic model, general dimorphic model, Gaussian plot model