Answers are grounded in this paper and cite their sources.
This work employs deep learning to classify tabla bols, indicating potential for automated transcription tools.
Tabla is an important percussion instrument in Indian classical music and is characterized by its unique strokes or a combination of strokes known as “bols.” Different bols are generated by striking the pair of tabla drums at various locations. Accurate identification of these bols is essential for automatic tabla transcription. In this work, we have employed a deep learning-based approach for automatic tabla bols classification. We implement and compare different architectures: a Convolutional Neural Network (CNN) and a Convolutional Recurrent Neural Network (CRNN). For the CRNN, we use two variations of RNN layers: GRU and LSTM. The Mel spectrogram is used as the input feature for the CNN model, while MFCC is used for the CRNN model. Various kinds of audio data augmentation techniques are employed to improve model accuracy. Our dataset consists of eight types of bols, which thirteen professional tabla players produce on two different tabla sets. Experimental results show that these methods achieve high classification accuracy and can contribute to developing automated tools for Indian classical music instruments.
No takes yet. Share an insight, caveat, or question.
Kumar et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: