Skip to content

AI Song Checker

Check whether a song was generated using AI — free, instant and private.

Paste nothing, register nothing, pay nothing. Choose an audio file, and this AI song checker measures the recording on your own device and reports an estimated AI probability with the reasoning attached.

It supports material from Suno, Udio, ElevenLabs Music, Stable Audio, Riffusion and similar tools — though, importantly, it estimates whether audio looks generated rather than naming which platform generated it.

Upload Audio or Drag & Drop

Upload an audio file to receive a probabilistic analysis of characteristics associated with AI-generated music.

  • MP3
  • WAV
  • M4A
  • FLAC
  • OGG
  • AAC
  • WebM

MP3, WAV, FLAC, AAC, M4A, OGG, WebM · max 25 MB · min 10 seconds · 30+ seconds recommended

Your audio never leaves your device. Decoding and analysis run entirely in this browser tab, and nothing is uploaded to a server. Your audio is processed only to perform this analysis, and your uploaded audio and temporary analysis data are automatically deleted after processing. No report links are created, and your analysis is never publicly accessible. Upload only audio you are authorised to process — analysis does not transfer ownership or publishing rights.

  • Free
  • Fast
  • Secure
  • No registration

How to check a song

The workflow takes under a minute. Drag an MP3, WAV, M4A, FLAC or OGG file onto the panel above, or pick it from your device. Your browser decodes the audio locally, samples several sections spread across the track, and measures each one.

You then get a report rather than a label: an estimated AI probability with a plausible range around it, a confidence level, a plain-language explanation of every measurement that contributed, and an explicit statement of what the result cannot establish.

  • Upload a file of at least ten seconds — longer is materially better
  • Prefer the highest-quality version you have; a 320 kbps MP3 or a WAV beats a 96 kbps rip
  • Avoid clips that are mostly silence, speech, applause or crowd noise
  • Check more than one section of a long track if the first result is inconclusive

Reading the result properly

The single most common mistake is reading a percentage as a verdict. A 78% estimate does not mean the track is 78% AI, nor that there is a 78% chance it is AI in any calibrated statistical sense. It means the measured signals lean that way by roughly that much, given an analysis engine that has not yet been validated against a labelled dataset.

That is why every result carries a range instead of a bare number, and why the checker caps its own output between 15% and 85%. A tool that returned 99% would be claiming a precision no acoustic detector currently has.

Confidence is a separate axis and matters just as much. A high probability with low confidence — usually because the sampled sections disagreed, or the file was heavily compressed — is much weaker evidence than a moderate probability with high confidence.

What the checker cannot tell you

It cannot name the generator. It cannot separate AI vocals from AI instrumentals. It cannot prove authorship, and it is not forensic evidence. It cannot see through heavy remixing, re-recording or aggressive mastering, and it cannot recognise a specific song or artist because it holds no database at all.

It is also asymmetric: much better at raising suspicion than at clearing a track. A 'likely human-created' result is the absence of a signal, not a certificate of authenticity.

Privacy

Nothing is uploaded. Decoding and analysis both run inside your browser tab using the Web Audio API, so no copy of your music reaches our servers. Deletion is automatic in the strongest sense — the decoded audio lives in memory only for the duration of the analysis and is discarded when you remove the file or close the tab.

There is no account, so there is no history of what you checked and no profile attached to it.

Frequently asked questions

  • The audio signal, and nothing else. It measures where the frequency content stops, how consistent the tonal balance is between sections, how much dynamic range survives, how much the brightness moves and how the stereo channels behave. It does not check a database, a fingerprint, a metadata field or a rights registry.