Speech span
Comparable speech is present in the selected interval.
For you / Content checks
It sounds familiar.Is the request genuine?
A voice can sound like someone you know. That alone is not a reason to send money or share information.
See how it worksVoice Authenticity
In plain English.
Voice authenticity asks whether a recording may have been generated or altered. It is a different question from whether two recordings sound like the same speaker.
A situation
worth a second look.
A familiar-sounding caller asks you to approve an urgent payment.
Defeated / content checkGuided example
“Can you send that payment now?”
Illustrative transcript · no audio analysisurgent-request.m4a
See what a useful answer looks like.
A familiar voice. An unusual request.
A signal, the reason behind it, and a practical next step. This guided example uses fixed sample findings, not a live detection model.
A clearer next step.
Pause. Call back on a number you already know before acting on the request.
Inspect your own contentFor technical teams: explore the method
Illustrative product views and fixed sample findings. Advanced analysis requires approved services and suitable reference data.
00:04.2CHANNEL + SYNTHESIS LANES
Channel artifacts limit the comparison; request the original recording.
Voice evidence does not prove who authored a message, who controlled a device, or whether a statement is true.
How the review works
1Preserve audio
Hash the exact recording and retain container details.
2Select spans
Exclude silence, overlap, and low-quality regions.
3Separate lanes
Keep speaker, synthesis, splice, and channel evidence distinct.
4State limits
Explain what another recording could resolve.