<p>Coastal areas have historically birthed numerous civilizations. One of the main forcing factors in these regions is spectral waves. Predicting wave characteristics can help protect coastal and marine structures from damage. Recently, several hybrid networks have been proposed to enhance the accuracy of wave characteristics predictions. This study developed a new hybrid network by means of Backpropagation Neural Network (BNN) and Discrete Wavelet Transform (DWT) in a serial configuration. By applying various and flexible functions in DWT, it is anticipated that BNN accuracy would be enhanced. This hypothesis was examined by a dataset from the Busher Port buoy, located in the Persian Gulf (consisted of wind and wave data). The dataset was split into 80% for training, 10% for validation, and 10% for testing to ensure robust model development and evaluation. The initial assessment revealed that BNN offered mediocre accuracy in predicting wave characteristics, yielding R² values of 0.788 for wave direction, 0.077 for wave period, and 0.0277 for wave height. Nevertheless, usage of DWT indicated that the accuracy of DWT-BNN relies on filter quantity and is relatively unaffected by the choice of mother function. It was also observed that symmetric functions were ineffective due to the irregular nature of waves. With optimized wavelet configurations, DWT-BNN significantly enhanced R² to 0.848 for wave direction, 0.681 for wave period, and 0.847 for wave height.</p>