Аннотация:We consider the problem of restoration of a signal that is measured in different frequency ranges by a set of spectrometers. The main goal is a demonstration of the possibility to restore the signal on the base of such scheme of registration and then to identify the medium. This problem is close to the registration of acoustic signals and also other types of signals. We suggest a new method for multichannel registration and processing of a noisy regular acoustic signal by the set of integral measurements, made simultaneously in each frequency range. This technique allows the reconstruction of partial signals in each frequency range, then sums them up and to obtain the full acoustic signal, containing both noisy and the regular part. As it shows one can to define the signal even for the case when the noise level is ten times more intense than the regular signal. Further analysis of the spectral dynamics of the full acoustic signal allows us to separate the regular part of the signal from its noisy part. The computer modeling we have carried out demonstrates many advantages of this algorithm. It provides a possibility to obtain the complete information about the dynamics of many spectral lines simultaneously and so to determine the spectral lines dynamics at regular frequencies. The proposed method also can be very useful for detection and identification of explosives in the enhanced GHz-THz range.