The paper analyses spectral and time-frequency characteristics of hydroacoustic signals of animal and anthropogenic origin, as well as background signals. The study aims to classify and identify these signals to address ecological monitoring tasks in the marine environment and to develop effective criteria for signal differentiation for automated assessment of the acoustic situation in coastal and shelf zones. We used methods of spectral and time-frequency analysis along with comparative analysis based on a review of current scientific literature. Characteristic features of spectra and spectrograms for various groups of signal sources were identified. Signals were classified according to their acoustic origin, and key parameters for signal identification under high noise conditions were determined, including spectral shapes, presence of harmonics, pulse durations, and specific temporal patterns. A feature set in the form of numerical vectors was created for subsequent application in machine learning algorithms and automatic recognition systems. The developed approach can be integrated into ecological monitoring systems for coastal waters and advanced navigation solutions.