Real-time audio deepfake and synthetic media detection with invisible content watermarking

Axory AI was founded by Tarini Sai Padmanabhuni after a friend with a large online following was targeted by a malicious deepfake and found that authorities would not act without technical evidence of manipulation. That gap, between an obvious harm and anything an investigator can use, is what the detection product addresses. The service analyses audio, voice, images, video, and documents for signs of synthetic generation or tampering, returns a verdict with a confidence score in real time rather than in batch, and is delivered as an API and SDK so it can sit inside an existing call flow, KYC check, or content pipeline instead of as a separate console.

The second product is invisible watermarking, which approaches authenticity from the other direction. Instead of judging an unknown file after the fact, it embeds a unique, imperceptible fingerprint into audio and video, including injection into live streams and calls, so provenance can be verified later. The company says the marks are built to survive compression, re-encoding, editing, and adversarial attempts at removal, and can be extracted to trace where content was leaked, cloned, or redistributed. Axory AI also sells callerAgent, an outbound AI voice agent for lead qualification and demo booking, which shares the underlying speech research. The company holds startup support from NVIDIA Inception, AWS Activate, and ElevenLabs Grants.