Search papers, labs, and topics across Lattice.
This study investigates the robustness of AI-generated music detectors when faced with generator sources not included in their training, using a novel evaluation framework called CoMoE to analyze audio token spaces. By conducting source-restricted evaluations on the \textsc{MoM-open} dataset, the authors reveal that while standard splits show saturation, significant disparities emerge under fake-source restrictions, indicating that different token spaces perform variably depending on the training source. The findings highlight the importance of considering codec-style discrete token spaces as a critical factor in AI-generated music detection, particularly in the context of generator shifts.
AI-generated music detectors reveal hidden vulnerabilities when tested against unseen generator sources, with token performance varying dramatically based on training data.
AI-generated music detectors can appear robust on standard benchmark splits, yet their deployments require transfer to generator sources absent during training. We study this problem with source-restricted evaluation on \textsc{MoM-open}, an open reconstruction of MoM-CLAM that replaces the non-redistributable real corpus with FMA and MTG-Jamendo while preserving the fake-generator protocol. To isolate the role of representation, we introduce \textsc{CoMoE}, a compact fixed classifier for comparing heterogeneous audio token spaces while keeping the downstream architecture and training recipe unchanged. Experiments show that standard and real-source-restricted splits are nearly saturated, whereas fake-source restriction exposes large differences between token spaces: X-Codec tokens are strongest when training on Udio alone, while MERT-derived tokens are stronger when training on Suno-v3.5 alone. These results suggest that codec-style discrete token spaces should be treated as a primary experimental axis under generator shift in AI-generated music detection. Our code and data are available at https://github.com/MAAP-LAB/CoMoE.
