Similar speaker recognition method and system using a non-linear analysis

The present invention relates to a similar-speaker recognition method and system using a non-linear analysis. An object of the present invention is to solve the similar-speaker recognition problem through a non-linear analysis of the audio signal to extract the non-linear characteristics present in the speech signal, and through a combination of the linear feature, such as spectrum. The present invention includes the steps of converting the voice data in a time space as the state vector on the phase space to characterized by using a non-linear characteristic of the speech to the speaker identification, extracts the information of the non-linear voice and; It consists of applying a non-linear time series analysis method to express non-linear characteristics of the reconstructed state vector. According to the present invention, it is possible to overcome the technical limitations of existing alignment algorithm, the technical impact of the audio-related application systems other than the speaker recognition system. Speaker identification, non-linear analysis, a linear analysis, the audio signal, the state vector, formants