ArtifactBench: Evaluating AI Music Detectors Under Distribution Shift

A paper submitted to arXiv on 20 Sep 2026 introduces ArtifactBench, a lineage-aware evaluation suite for AI-generated music detectors that separates calibration from final testing and reports source-level performance with uncertainty. On the 562-track common-success test intersection, ArtifactNet reaches 0.982 AUROC and 0.918 balanced accuracy, versus 0.761 AUROC and 0.776 balanced accuracy for the public Deezer detector, while SpecTTTra and CLAM fall below 0.30 AUROC under the shifted cohort. The authors report that leakage control, cohort availability, threshold policy, and model-specific missingness alter measured performance and model rankings, exposing generator- and real-domain shifts that aggregate scores conceal.

Computer Science Sound Submitted on 20 Sep 2026 Title:ArtifactBench: Lineage-Aware Evaluation of AI-Generated Music Detectors under Distribution Shift View PDF https://arxiv.org/pdf/2609.23550 HTML experimental https://arxiv.org/html/2609.23550v1 Abstract:AI-generated music detectors are commonly compared using aggregate scores on benchmarks whose training overlap, generator lineage, source provenance, and audio-transformation history are only partially observable. This paper introduces ArtifactBench, a lineage-aware evaluation suite for measuring detector behavior across generator families and versions, real-music domains, collection-cohort shift, and inference coverage. The benchmark groups source recordings and their derived variants by content identity, separates calibration from final testing, records inference failures independently from classification errors, and reports source-level performance with uncertainty in addition to aggregate metrics. We evaluate multiple publicly available detectors under a version-pinned common protocol and examine how leakage control, cohort availability, threshold policy, and model-specific missingness alter measured performance and model ranking. On the 562-track common-success test intersection, ArtifactNet obtains 0.982 AUROC and 0.918 balanced accuracy, compared with 0.761/0.776 for the public Deezer detector; SpecTTTra and CLAM fall below 0.30 AUROC under this shifted cohort. These results also expose substantial generator- and real-domain shifts that aggregate scores alone conceal. Current browse context: cs.SD References & Citations Loading... Bibliographic and Citation Tools Bibliographic Explorer What is the Explorer? https://info.arxiv.org/labs/showcase.html arxiv-bibliographic-explorer Connected Papers What is Connected Papers? https://www.connectedpapers.com/about Litmaps What is Litmaps? https://www.litmaps.co/ scite Smart Citations What are Smart Citations? https://www.scite.ai/ Code, Data and Media Associated with this Article alphaXiv What is alphaXiv? https://alphaxiv.org/ CatalyzeX Code Finder for Papers What is CatalyzeX? https://www.catalyzex.com DagsHub What is DagsHub? https://dagshub.com/ Gotit.pub What is GotitPub? http://gotit.pub/faq Hugging Face What is Huggingface? https://huggingface.co/huggingface ScienceCast What is ScienceCast? https://sciencecast.org/welcome Demos Recommenders and Search Tools Influence Flower What are Influence Flowers? https://influencemap.cmlab.dev/ CORE Recommender What is CORE? https://core.ac.uk/services/recommender arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs https://info.arxiv.org/labs/index.html .