Accurate prediction of airfoil self-noise is crucial for the design of quiet Urban Air Mobility (UAM) vehicles, as conventional methods like wind tunnel testing and Computational Fluid Dynamics (CFD) are often high-cost or computationally intensive. This study proposes a reliable and fast hybrid prediction framework for the self-noise estimation of the NACA 0012 airfoil based on experimental data. The proposed methodology utilizes the Adaptive Network Based Fuzzy Inference System (ANFIS), with its membership function parameters optimized using three distinct metaheuristic algorithms: Particle Swarm Optimization (PSO), Genetic Algorithm (GA), and Ant Colony Optimization (ACO). The models were trained and validated using open-source experimental datasets from NASA. Comparative analysis revealed that the PSO-ANFIS hybrid model demonstrated the most superior performance, achieving a minimal Root Mean Square Error (RMSE) of 2.6684 on the training data and 2.8521 on the independent test data. The close proximity of the training and testing error metrics confirms the model’s high accuracy and excellent generalization capability without overfitting. This successful integration of metaheuristics and ANFIS provides a reliable, fast, and lowcost tool for complex aerodynamic noise prediction and can serve as a starting point for the design of low-noise airfoils.