In this paper the possibilities of modern audioprocessing systems have been described. One of the most important tasks is a proper definition of feature v ectors. For example, the long-term spectra vector has been chos en a the background for text-independent verification sy stem. In this case it is possible to compare specific featur es of the current speaker voice with the computer base contai ni g stored vectors of known speakers. As the results, t he distances between the vector of unknown person and the vectors of the persons included in the base have be en obtained. Afterwards, the possibility of hardware realization of some voice processing components has been discussed.