The proliferation of audio data poses significant challenges in efficiently analyzing and classifying large datasets. Using the MobileNetV2 architecture, which is known for its efficiency and effectiveness in handling complex data, this study addresses the task of classification of genres in audio data. Leveraging MobileNetV2, we preprocess the audio data by segmenting it into homogeneous content and extract features using various techniques. These features are then fed into MobileNetV2 model, where they undergo hierarchical representation learning. We experiment with different model architectures, including combinations with Recurrent Neural Networks (RNNs), to evaluate performance. Our results demonstrate the efficacy of MobileNetV2 in achieving high classification accuracy, with an overall accuracy of 90.3%. This study underscores the utility of MobileNetV2 in efficiently analyzing and classifying large volumes of audio data, with implications for applications such as music streaming services and recommendation systems.