Music Genre Classification Using Long Short-Term Memory (LSTM) Networks
Suman Kumar Swarnkar, Yogesh Kumar Rathore
Abstract
This examination explores music-type characterization utilizing long short-term memory (LSTM) networks applied to sound spectrograms, intending to upgrade interactive media understanding. The procedure includes information preprocessing, highlight extraction, LSTM network preparation, and assessment. Tests show the viability of the proposed approach, accomplishing a characterization exactness of 85%. Correlation with pattern techniques uncovers predominant execution measurements, with accuracy, review, and F1 score upsides of 0.86, 0.85, and 0.85, respectively. The LSTM organization’s capacity to catch worldly conditions and perceive complex examples in music information adds to its strength and adaptability. Moreover, the examination features the significance of profound learning methods in tending to difficulties in music kind arrangement and related errands.