Time-Specific Classification in Hindustani Classical Music
Priya Mishra, Manish Kalra · 2024
This study presents a novel approach to classify morning and evening raags in Indian classical music using a unique architecture named the “1d-SRiyam” network. The classification is based on time-specific characteristics, capturing the emotive expressions and melodic nuances inherent in each raag. The proposed model was compared with a conventional 1d CNN-LSTM model using evaluation metrics, including accuracy, precision, recall, and F1 scores. Results show that the 1d-SRiyam network outperforms the conventional model with an accuracy of 75% compared to 59%. The precision for classifying morning raags (class 0) was 70%, while for evening raags (class 1), it was 76%. The recall for class 0 was 38% and for class 1, it was 92%. The F1 scores were 0.49 for class 0 and 0.83 for class 1, indicating the superior predictive capabilities of the proposed model. The findings of this research highlight the importance of considering time theory in raag classification and provide valuable insights into the emotional and cultural aspects of Indian classical music. The 1d-SRiyam network offers a promising avenue for further exploration and appreciation of the timeless artistry present in morning and evening raags, enriching the understanding of this classical musical heritage.