Korea AI Forensic Labs analyzes audio for signals of AI generation and manipulation, and delivers precision measurement reports prepared for courts, investigators, and rights organizations.
OPEN ANALYSIS CONSOLE HOW WE WORKA questioned recording is compared against verified reference recordings of the speaker. We measure synthetic-speech signals, identity consistency, and traces of editing, and localize every finding on the timeline. Designed for fraud investigations, impersonation disputes, and contested recordings.
A full track is analyzed for AI-generation signals, section by section and instrument by instrument. Where AI signals are observed, further signature analysis narrows their likely origin. Designed for copyright societies, distributors, and platforms facing AI-generated submissions.
Every analysis is delivered as a structured measurement report: cryptographic file integrity, methodology, measured values, time-localized findings, and the limitations of each measurement, stated plainly. Prepared for submission to courts and investigative agencies.
Our reports state which signals were detected, where, and at what measured strength, together with the supporting evidence. Whether a recording is “fake” is a determination that rests with courts and investigators, and we believe that is exactly where it should rest.
A measurement that cannot support a conclusion is reported as such. We consider a disciplined “inconclusive” more valuable to justice than a confident guess.
Each measured value is reported within its verified operating range. Nothing is hidden to make a finding look stronger than it is.
Designed for courts and legal counsel, law enforcement, copyright societies, music distributors, platforms, and broadcasters.
Our analysis begins with sound. Image and video forensics are in development.