ITRC Exhibition

Anti-Cloning

Proactive protection System

Technology to protect voices from audio deepfakes

Technology Overview

Why is Anti-Cloning Technology Needed?

With advances in deep learning, it has become possible to precisely replicate
anyone's voice from just a short audio sample.

01
🎭
Voice Cloning Threat
Modern AI voice synthesis can realistically reproduce a person's voice from just a short audio sample.
Such technology is increasingly being exploited for voice phishing and unauthorized content creation.
02
⚡
Cloning Attack Method
Attackers input publicly available voice data into TTS models to easily generate fake voices that sound like the original speaker. This poses a very high risk as anyone can attempt it without special equipment.
03
🛡️
Our Technology's Solution
We embed imperceptible protection signals into audio. When TTS is trained on protected audio, the synthesizer fails to reproduce the speaker's voice.

Generation Suppression Pipeline

🎙️ Original Voice Speaker's actual voice
🛡️ Protection Applied Inaudible signal embedded
🔊 Protected Voice Same listening quality
🤖 Clone Attempt TTS model training
❌ Clone Failed Speaker voice unrecoverable
Audio Demo

Original Voice vs Protected Voice

Even with protection signals embedded, there is little perceptible difference to human ears. Listen for yourself.

Select Speaker
TTS Demo

Original TTS vs Protected TTS

TTS trained on protected audio has its cloning neutralized, significantly reducing voice similarity.

Select Speaker
Comparison

Compare All 4 Voices at a Glance

From original to failed clone — experience the effect of protection technology firsthand.

Select Speaker