Real-time Music Genre Classification Using Cnn and Mfcc for Interactive Systems
Year: 2026
Type: article
Abstract: Real-time music genre classification is an essential task in Music Information Retrieval (MIR), particularly for interactive systems that require both high classification accuracy and low-latency response. Many existing deep learning approaches focus primarily on offline accuracy, which often results in increased computational complexity and limited suitability for real-time deployment. This research addresses the problem of accurately classifying music genres from high-dimensional audio signals under real-time constraints. The objective of this study is to develop an efficient and deployable classification framework that balances predictive performance with computational efficiency. It is hypothesized that Mel-Frequency Cepstral Coefficients (MFCCs), when combined with a lightweight and regularized Convolutional Neural Network (CNN), can effectively reduce feature dimensionality while preserving discriminative information. The proposed approach employs MFCCbased feature extraction followed by a CNN architecture optimized using regularization and data augmentation techniques. Experimental evaluation is conducted using the GTZAN benchmark dataset consisting of ten music genres. To validate real-time applicability, the trained model is deployed within an interactive Streamlit-based application supporting live audio input. The results demonstrate that the proposed MFCC-CNN framework achieves a classification accuracy of up to 95% while maintaining low inference latency suitable for interactive systems. The findings confirm that the proposed approach effectively addresses the accuracy-latency trade-off in real-time music genre classification. Keywords: Convolutional Neural Networks, Interactive Systems, Mel-Frequency Cepstral Coefficients, Music Genre Classification, Real-Time Audio Processing.
Language: English
Topic: Music and Audio Processing
Subfield: Signal Processing
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7162215270