Urban areas face persistent environmental noise, adversely impacting human health and well-being. Automatic acoustic noise classification can assist in monitoring and mitigating this problem. This paper introduces a technique for environmental acoustic noise classification, the proposed approach involves three stages (i) Extraction of Mel-frequency Cepstral Coefficients (MFCCs) from acoustic signals, which captures the spectral characteristics of the noise, (ii) Employing a CNN classifier trained on the MFCC feature vectors of the UrbanSound8K dataset, and (iii) Extraction of the environmental noise from recorded audio using Fast Fourier Transform (FFT) and Inverse Fast Fourier Transform (IFFT). Then, the MFCC features of the extracted noise are used to classify it into one of the ten predefined classes using the CNN model. The overall test accuracy for classifying noise (embedded in audio recordings) is 90.44%, which indicates that our model has achieved a commendable level of accuracy. Our work offers a fresh perspective on the problem and proposes a simple and novel solution.