Last updated: July 14, 2026
When you run a scan, your audio is sent over an encrypted connection to our AI detection system, which analyzes it to estimate the likelihood it was AI-generated or AI-assisted. We do not store your uploaded audio files on our own servers, and your audio is never used to train AI models and is never sold or shared for marketing. We keep only your analysis results (such as the score and the file name) in your account, and you can remove them any time using the "Delete my data" option in the results view.
No. We do not store your uploaded audio on our servers and we do not use it to train any AI model. Your audio is sent to our AI detection system only to produce your result. We keep the analysis results in your account until you delete them, and you keep all rights to your music.
When you upload a track, our system extracts an acoustic fingerprint — a mathematical map of the audio's spectral and timing patterns. That fingerprint is then analyzed by our AI detection system, which looks for characteristics that separate human recordings from AI-generated audio:
Our system scores each track across these dimensions and returns a likelihood score — the probability that the track is AI-generated.
Each scan returns two types of signals:
Use the binary result for straightforward decisions. Use the probability score as a confidence reference. Consider results alongside other contextual signals where appropriate.
Our AI detection system is built on state-of-the-art models evaluated against large internal test datasets. Performance figures from our current model version:
| Metric | Result |
|---|---|
| Precision (AI) | 99.98% |
| Recall (AI) | 99.94% |
| False Positive Rate | 0.06% of human tracks |
| False Negative Rate | 0.06% of AI tracks |
Per-source accuracy breakdown:
| Source | Accuracy |
|---|---|
| Human recordings | 99.91% |
| Suno | 99.86% |
| Udio | 99.93% |
| Sonauto | 98.81% |
| Mureka | 99.99% |
| Riffusion | 98.00% |
These figures are based on internal evaluation data. Actual performance may vary depending on audio quality, compression, mixing, editing, and real-world usage conditions.
Our detection system currently identifies content consistent with the following AI music sources:
We are actively expanding coverage and working to include additional sources such as ElevenLabs, MiniMax, and Seed-Music. Supported source coverage may evolve over time as generative audio models develop.
AISongScan provides a probabilistic assessment, not a definitive verdict. It gives you a likelihood signal based on audio characteristics; it does not prove that a track is human-made or AI-generated. Results should be treated as one signal among many, not as legal or evidentiary proof.
AISongScan accepts common audio formats such as MP3, WAV, M4A, and FLAC. For best results, upload a clear, full-length or near-full-length track rather than a short clip.
Yes. Scanning requires a free account so we can apply your scan allowance fairly. Sign-in is passwordless: enter your email, and we send you a secure single-use link. There is no password to remember.
Every new account includes a set of free scans to try the Service. If you need more, paid plans (Creator and Professional) offer higher monthly scan allowances and are available on monthly or annual billing. You can view current plans and pricing from the upgrade option inside the app.
Open the account menu in the app and choose to manage your subscription. This opens a secure billing portal where you can update your payment method, switch plans, or cancel. If you cancel, you keep access until the end of your current paid period.
Use the "Delete my data" option in the results interface to remove a scan's results at any time. We do not store your uploaded audio on our servers in the first place. To close your account or request removal of your account information, contact us at the address below.
Yes. Payments are handled by Stripe, a leading payment processor. Your card details are entered directly with Stripe and are never stored on our servers.
Detection results are probabilistic model outputs based on patterns identified in submitted audio content. Results are intended to assist review and inspection workflows and should not be interpreted as definitive proof of content origin. Performance metrics are based on internal evaluation data and may vary in real-world use cases. Detection results should not be used as the sole basis for legal, enforcement, or other high-impact decisions without additional review.
We are happy to help. Email us at contact@stablestateit.com and we will get back to you.
StableState LLC
Wisconsin, United States