Spectral analysis
Looks for time-frequency textures, codec or decoder artefacts and generator-like patterns in spectrogram representations.
Shadermic Tools
AI Song Detector
Upload a song to estimate whether it contains patterns commonly associated with AI-generated music. Results are probabilistic and may be inconclusive.
MP3, WAV, FLAC, M4APrivate account accessUncertainty-aware report
Account required
Sign up through the dashboard to analyze songs.
Your free Shadermic account protects the upload endpoint, keeps anonymous abuse away and lets us add analysis history later.
- Secure member-only upload
- No public anonymous processing
- AI, human or inconclusive result states
How the report is designed
Multiple signals, with room for uncertainty.
A responsible detector should not force every song into a definitive yes/no decision. The connected model can combine several branches and return an inconclusive state when evidence is weak.
Temporal analysis
Reviews how rhythm, melody, harmony and production structure evolve across longer sections of the song.
Calibrated result
Returns likely AI, likely human or inconclusive, together with branch scores and a visible limitation notice.
Privacy and limits
The plugin does not save the raw upload to the WordPress Media Library. The configured detector backend must use minimal retention. Results are forensic estimates and must not be treated as proof, legal certification or the sole basis for punitive action.