AI Voice Detector
AI Voice Detector checks a voice recording and tells you whether the speech was made by AI. You upload a voice note or call recording and get back a verdict with an AI probability score. A breakdown shows which parts of the recording sound synthetic.
Journalists use it to check suspicious audio before publication, and investigators keep a dated record of every check. Banks and call centers screen a flagged call before a payment goes out. Families use it on a voicemail from a relative asking for money.
How a check works
Drag in an audio file with at least 3 seconds of clear speech. It reads MP3, WAV, M4A, OGG, FLAC and AAC files, and video files too. The model scores the clip segment by segment and returns a verdict on a four-step scale: AI Generated, Likely AI, Likely Human or Human. Longer clips show which segments were flagged, so you can see if only part of a recording was synthetic.
You don't need an account for a free check. Sign in when you want to save results to your history or share them with a link that doesn't expire.
For teams and developers
Paid plans add detailed confidence scores and PDF or CSV export of your history. They also include a REST API that returns the verdict and the segment scores as JSON. The API uses the same model as the web detector. A Chrome extension for checking audio from any browser tab is in development.
Pricing
AI Voice Detector has a free plan with 50 checks a month on clips up to two minutes, with no card required. Paid plans raise the monthly limit and allow clips up to 10 minutes, and they add export and API access. Enterprise volumes are available on request.
Can AI Voice Detector prove a recording is fake?
It gives a probability, not proof. Use the verdict and segment scores alongside other evidence, such as where the recording came from and who sent it.

TRY IT