Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{188745,
        author = {Anand Sarode and Deep Hatwar and Tejas Hundare and Sarvadnya Kankhare and Prof. Sagar Apune},
        title = {Noctune: AI-Powered Music Playlist Generator},
        journal = {International Journal of Innovative Research in Technology},
        year = {2025},
        volume = {12},
        number = {7},
        pages = {3583-3591},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=188745},
        abstract = {The rapid expansion of digital music platforms has resulted in enormous audio libraries containing millions of tracks, significantly complicating the process of discovering music that aligns with a listener’s preferences. Traditional recommender systems rely heavily on collaborative filtering or metadata analysis, both of which fail to capture the acoustic qualities that fundamentally determine perceptual similarity between songs. This research proposes an intelligent, audio- driven playlist generator that leverages a comprehensive set of audio features extracted from the Free Music Archive (FMA) dataset. By incorporating Mel-Frequency Cepstral Coefficients (MFCCs), Chroma features, Spectral descriptors, and rhythm- related attributes, the system constructs a high dimensional representation of each song’s acoustic identity. These representations are standardized and processed through a K-Nearest Neighbors (KNN) model to identify similar tracks in the dataset. The system is deployed through a Flask-based web application that allows users to upload audio files in real time, automatically extract their features, and obtain a playlist consisting of acoustically similar songs. Experimental evaluation demonstrates high recommendation accuracy, consistent relevance across diverse genres, and fast response times suitable for real-world usage. The findings affirm that audio-based recommendation approaches offer significant advantages over meta data driven systems, especially in scenarios involving new, unlabeled, or obscure music. The implications of this work extend to music discovery platforms, educational tools, and musicological research.},
        keywords = {Music Recommendation, MFCC, KNN, Audio Features, Music Information Retrieval, Content-Based Filtering},
        month = {December},
        }