"Researchers Explore Continuous Liveness Detection for Voice Biometrics"

A new study by scientists from Florida State University and Rutgers University delved into the possibility of continuous liveness detection for voice biometrics implemented on smart devices. They released a paper with the goal of strengthening security for voice biometrics as this technology is claimed to be vulnerable to spoofing through replay attacks. These attacks involve using pre-recorded voice biometric samples from a genuine user to spoof voice authentication systems. The researchers designed a liveness detection tool called VoiceGesture for smart devices such as smartphones and smart speakers. It is capable of differentiating a live user from a recording by using both the user's unique articulatory gesture when they say a passphrase and smart devices' audio hardware advances. According to the researchers, the VoiceGesture tool does not require specialized hardware since it works via a speaker and a microphone commonly implemented on smart devices that support voice input. Testing showed that VoiceGesture achieves more than 99 percent detection accuracy for text-dependent liveness detection and around 98 percent for liveness detection independent of text. This article continues to discuss the liveness detection tool VoiceGesture designed to establish a new level of security for voice biometrics.

Biometric Update reports "Researchers Explore Continuous Liveness Detection for Voice Biometrics"