Search papers, labs, and topics across Lattice.
This study investigates the detection of AI-generated stems within hybrid human-AI music tracks, marking the first exploration of this domain. By utilizing the MUSDB18-HQ database and comparing detection strategies, the authors find that traditional source separation methods fail to recover AI-related artifacts effectively. They propose a novel parallel architecture that leverages source-relative energy estimates and simple binary classifiers, achieving promising results in identifying AI-generated components in mixed audio tracks.
AI-generated music stems can be detected even when mixed with human-produced audio, challenging the limits of current detection methods.
This paper presents, to the best of our knowledge, the first study on detecting human-AI hybrid music tracks created by mixing human-produced and AI-generated stems. Building on recent work showing that AI music detectors can identify decoder-related artifacts in fully generated music, we investigate whether such artifacts remain detectable at the stem level after mixing. Using MUSDB18-HQ database in a two-stem vocals + accompaniment setting, we simulate hybrid mixtures by autoencoding individual stems with a neural codec. We compare two strategies combining AI-generated mix detection and source separation. A naive sequential pipeline, where source separation is followed by detection on separated sources, confirms that artifacts associated with an AI-generated stem are not reliably recovered by generic source separation systems. We therefore propose a parallel architecture in which source separation is only used to estimate source-relative energy within the mixture. We then train simple stem-specific binary classifiers that take as input the generated mix prediction together with the relative energy of the target stem on short audio chunks. Averaging chunk-level predictions yields encouraging track-level results, highlighting the potential of such approaches for detecting AI-generated stems in hybrid music.
