# ArtifactNet: Detecting AI-Generated Music via Forensic Residual Physics

Source: [ArtifactNet: Detecting AI-Generated Music via Forensic Residual Physics](<https://www.sota2.com/research/paper/69e5b32d7d0172140ec7e165>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-04-23T06:00:02.332Z

Authors: Heewon Oh

Published: 2026-04-17T17:14:07.000Z

## Abstract

We present ArtifactNet, a lightweight framework that detects AI-generated music by reframing the problem as forensic physics -- extracting and analyzing the physical artifacts that neural audio codecs inevitably imprint on generated audio. A bounded-mask UNet (ArtifactUNet, 3.6M parameters) extracts codec residuals from magnitude spectrograms, which are then decomposed via HPSS into 7-channel forensic features for classification by a compact CNN (0.4M parameters; 4.0M total). We introduce ArtifactBench, a multi-generator evaluation benchmark comprising 6,183 tracks (4,383 AI from 22 generators and 1,800 real from 6 diverse sources). Each track is tagged with bench\_origin for fair zero-shot evaluation. On the unseen test partition (n=2,263), ArtifactNet achieves F1 = 0.9829 with FPR = 1.49%, compared to CLAM (F1 = 0.7576, FPR = 69.26%) and SpecTTTra (F1 = 0.7713, FPR = 19.43%) evaluated under identical conditions with published checkpoints. Codec-aware training (4-way WAV/MP3/AAC/Opus augmentation) further reduces cross-codec probability drift by 83% (Delta = 0.95 -\> 0.16), resolving the primary codec-invariance failure mode. These results establish forensic physics -- direct extraction of codec-level artifacts -- as a more generalizable and parameter-efficient paradigm for AI music detection than representation learning, using 49x fewer parameters than CLAM and 4.8x fewer than SpecTTTra.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [AI Music Detection on ArtifactBench (test)](<https://www.sota2.com/research/sota/ai-music-detection-on-artifactbench-test.md>): AI Music Detection; ArtifactBench (test); F1 Score: 98.29.

- [AI-generated audio detection on ArtifactBench v1 (test)](<https://www.sota2.com/research/sota/ai-generated-audio-detection-on-artifactbench-v1-test.md>): AI-generated audio detection; ArtifactBench v1 (test); Fail Count: 4.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2604.16254>)

[PDF](<https://arxiv.org/pdf/2604.16254.pdf>)

[Implementation](<https://github.com/Intrect-io/artifactbench>)