Rock properties are vital for applications in in mining, geotechnical engineering and various rock engineering projects. The P-wave velcoity is a key indicator for evaluating the integrity and stability of rock formations, critical for tasks like tunnel excavations, slope stabilization, and other mining operations. It also serves as an essential parameter in designing foundations for structures such as dams, bridges, and other rock-based constructions. The accurate determination of P-wave velocity depends on obtaining high-quality cored rock samples, but challenges like sample preparation, costs and time limitations have driven increased use of computational approaches for estimation. Earlier studies often relied on laboratory-based experiments and indirect techniques to estimate rock properties, including P-wave velocity. In the present study, a novel method is proposed to predict the P-wave velocity (Vp) of igneous rocks, specifically granite, by utilising the ball mill grinding characteristics which is an unconventional and reliable approach. A hybrid random forest model, enhanced through optimization with the dolphin swarm algorithm, is developed to estimate Vp based on grinding characteristics. The effectiveness of the model is evaluated in both training and testing stages using metrics such as the coefficient of determination (R2), Root Mean-Squared Error (RMSE), and Variance Accounted for (VAF), yielding values of 0.982, 168.09 m/s, and 98.22% for training, and 0.976, 138.83 m/s, and 97.7% for testing, respectively. This technique provides an efficient alternative to traditional P-wave velocity determination, eliminating the need for intricate sample preparation and expensive ultrasonic testing equipment.