Paper Contents
Abstract
Music Recommender System using a Spectrograms for a music content-based Librosa providing an efficient and recommendational system Audio feature extraction. User feedback In the age of internet music streaming, recommendation systems have become crucial to enhancing the experience of an individual by recommending songs based on previous songs listening history. The traditional models of making music recommendations rely more on collaborative filtering or content-based methods. However, these approaches enjoy some shortcomings such as the cold-start problem or feature limitations, where certain aspects of music cannot be incorporated. The objectives of this research include explaining advancements of a music recommender system based on the capabilities of Librosa, its software Python built for music and audio analysis, and the improvements of the recommendation together with the feature extraction based on the spectrogram. Leaving scope for the further research in the area of spectrograms as sound and image correlation adds weight to the scope of further research.The system uses Librosa to incorporate key features of audio tracks such as Mel-frequency cepstral coefficients, chroma and tempo values which represent the timbre, harmonic structure and rhythmic details of the track respectively. Besides that, the what spectrograms provide graphical representations of these parameters and were used in a graphical presentation of spectrotemporal tuning to the singer in order to the track elements of a the singer portraits.
Copyright
Copyright © 2025 Siddhant Godwani. This is an open access article distributed under the Creative Commons Attribution License.
