Automatic DJ Mix Generation Using Highlight Detection

Abstract

We demonstrate a fully automated system that generates DJ mixes from an arbitrary pool of audio tracks. Our goal is to offer a framework which automatically provides a high-quality endless mix of musical tracks without user supervision, given a seed track. To achieve this, we follow a filter and rank-based approach which selects the best-matching clip of a song to mix with the given seed track. We applied state-of-the-art musical information retrieval (MIR) methods including deep learning-based approaches such as automatic highlight extraction and key-tempo estimation for extracting cue points, choosing transition methods, and selecting audio clips. The results show that the proposed system can automatically generate competitive DJ mix contents plausible for users to enjoy.