Exploring CNN-Based Architectures for Vietnamese Traditional Music Genre Classification

With the rapid advancement of artificial intelligence, the classification of music genres is also progressing at a similar rate. Numerous corporations, including Spotify and Apple, have garnered significant interest in this domain, leading them to recruit highly skilled architects for their respective undertakings. Nevertheless, it is worth noting that various manifestations of Asian traditional music, including Vietnamese traditional music, have not attained an equivalent standard of performance excellence when compared to other musical genres. The objective of this study was to assess the effectiveness of Convolutional Neural Network (CNN) models in categorizing Vietnamese traditional music genres. In this study, we assessed the efficacy of two architectural designs, namely Convolutional Recurrent Neural Networks (CRNNs) and Parallel Convolutional Recurrent Neural Networks (PCRNNs), by employing diverse spectrograms. The present study involved a comparative analysis conducted on a well-maintained dataset of Vietnamese traditional music. The results of our study revealed the architecture that is most suited for this task. In our experiment, it has been proven that Parallel Convolutional Recurrent Neural Networks are the most suitable alternative for this specific purpose. This study makes a valuable contribution to the field of music information retrieval (MIR) by investigating the effectiveness and precision of Convolutional Neural Networks (CNN)-based approaches in classifying Vietnamese traditional music genres.