Zero-Shot Crate Digging: DJ Tool Retrieval Using Speech Activity, Music Structure And CLAP Embeddings

TL;DR

This paper presents a zero-shot audio retrieval system that uses speech activity detection, music structure analysis, and CLAP embeddings to identify DJ tools in personal music collections, aiding DJs in live and studio settings.

Contribution

It introduces a novel approach combining speech/music activity detection, music boundary analysis, and CLAP embeddings for zero-shot DJ tool retrieval.

Findings

Effective retrieval of DJ tools demonstrated in personal music collections.

System supports live and studio DJ performances.

Utilizes open-source tools for comprehensive audio analysis.

Abstract

In genres like Hip-Hop, RnB, Reggae, Dancehall and just about every Electronic/Dance/Club style, DJ tools are a special set of audio files curated to heighten the DJ's musical performance and creative mixing choices. In this work we demonstrate an approach to discovering DJ tools in personal music collections. Leveraging open-source libraries for speech/music activity, music boundary analysis and a Contrastive Language-Audio Pretraining (CLAP) model for zero-shot audio classification, we demonstrate a novel system designed to retrieve (or rediscover) compelling DJ tools for use live or in the studio.