In the field of artificial intelligence, achieving accurate environmental sound classification (ESC) is crucial for developing systems with robust acoustic awareness. Existing approaches often use limited feature sets, which can hinder performance in distinguishing subtle differences among sound classes. This paper addresses this gap by evaluating and aggregating various acoustic features, including Short-Time Fourier Transform (STFT), Spectral Centroid, Mel Spectrogram, Zero Crossing Rate, and Discrete Fourier Transform (DFT). Through comprehensive experimentation, we identified the best-performing feature combinations that optimize classification accuracy. Our comprehensive experiments demonstrate that combining STFT, Spectral Centroid, Mel Spectrogram, Zero Crossing Rate, and DFT significantly improves classification performance. By integrating these features, our solution effectively captures the intricate patterns and nuances of sound, addressing the shortcomings of earlier methods. The proposed approach outperforms existing models, showcasing its effectiveness in handling complex ESC tasks and offering robust improvements in performance.