In this paper, a technique for compressing data from a time-domain audio digital signal, using the Matrix Pencil Method (MPM) is presented. The results show that the selection of critical parameters like time window and sampling frequency can effectively work alongside MPM, to reduce an audio signal data to a set of residues and poles, and have an accurate reconstruction of the vector related to the original audio. The compression process is focused on acoustic signals coming from boat engines and the reconstruction made for these poles and residues provides an R <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> correlation coefficient near to 100%. Besides, is possible to obtain a rate of compression between 80 and 93%. The audio signals are collected by hydrophones deployed in a coastal area, close to the shore. The signal under process is expected to present frequency components up to 1KHz. The data reduction enhances the transfer time and the bandwidth of the wireless communication channel under maritime conditions. Also, the proposed system considers the case of multiple sensor locations, acquiring a signal from multiple points in the area of influence allowing higher-level processes as the identification and tracking of the observed boat in naval operations.