CrateDigger: Hybrid Acoustic DNA Recommendation Engine for Personalised Sequential Music Discovery

Read full text Summarise with VIRAT
Authors: Ovidio Danilo Pruneda Saldivar
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

We present CrateDigger, a music recommendation system designed for music discovery and sequential playlistconstruction that maps each track to a multi-dimensional normalised acoustic feature vector and applies a novelhybrid scoring engine to identify optimal track transitions. The system combines four independently weightedsignals: (1) acoustic cosine similarity via approximate nearest-neighbour search on a pgvector index, (2)personalised affinity derived from the user's historical interaction events with temporal decay, (3) obscurityweight inversely proportional to any normalised engagement or consumption frequency signal, and (4)contextual momentum incorporating session phase classification and tempo coherence. Additionally, the system implements a collective transition graph that aggregates real user behaviour acrossmultiple users — recording which tracks are played in sequence, with what harmonic compatibility (Camelotwheel), and with what implicit quality signal — and uses this graph to power a next-track suggestion algorithmthat combines collective wisdom with individual acoustic analysis.Adaptive presets learn each user's characteristic acoustic ranges for opening, peak, and closing phases byapplying percentile-based statistical analysis (P10–P90) over historical events, producing personalisedrecommendations that improve with continued use. The entire system runs on serverless edge infrastructure atzero marginal cost.

Fields of Study & Subjects

Recommender Systems and Techniques Music and Audio Processing Music Technology and Sound Studies

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
22 March 2026
Language
EN