Inter component phase processing of speech signals for their recognition and identification of announcers

The characteristics of the phase differences between quasi harmonic components of speech signals is proposed to use for the solution of speech recognition task and identification of the announcers. The simplest procedure of speech signals inter component phase processing is realized for harmonic on frequency of the basic tone and overtones with approximately divisible by its frequencies. The estimation of a phase difference between frequency components of analyzed signal can be distributed on formants as well using the coordinated moving of selective windows with multiple central frequencies or their rational attitude along a spectrum of a signal. The inter component phase processing is applicable not only for speech, but also for many other kinds of superbroadband and polyharmonic signals.