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

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.