Scholarlink

Music Genre Classification (MGC) is the process of classifying audio files into specific genres according to their acoustic features. This paper uses a combined approach of CNN and LSTM to improve the classification accuracy. The model was trained on the GTZAN dataset, which contains 1,000 30-second WAV audio tracks from 10 genres. Key preprocessing techniques, which included MFCC and spectral roll-off, were applied to extract important features. The hybrid model of the proposed CNN-LSTM achieved a classification accuracy of 96.68% while compared to conventional machine learning-based methods such as SVM and kNN. Therefore, by this analysis, the deep learning algorithm is well suited in applications related to music recommendation systems, education, and productions.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

GenreBeats: A Hybrid CNN-LSTM Approach for Music Genre Classification

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Jhansi Vazram Bolla,
  • Venkata Hari Pavan Kota,
  • Dola Sankar Sela,
  • Akash Segu,
  • Sireesha Kambhampati,
  • N. B. S Vijay Kumar

摘要

Music Genre Classification (MGC) is the process of classifying audio files into specific genres according to their acoustic features. This paper uses a combined approach of CNN and LSTM to improve the classification accuracy. The model was trained on the GTZAN dataset, which contains 1,000 30-second WAV audio tracks from 10 genres. Key preprocessing techniques, which included MFCC and spectral roll-off, were applied to extract important features. The hybrid model of the proposed CNN-LSTM achieved a classification accuracy of 96.68% while compared to conventional machine learning-based methods such as SVM and kNN. Therefore, by this analysis, the deep learning algorithm is well suited in applications related to music recommendation systems, education, and productions.