Voice Biometrics Group has developed a noise modeling system based on deep learning algorithms. Soufleris describes the challenges of using prior noise filtration techniques and why newer, learning-based algorithms are able to perform better by cutting EER by 40% or more. He discusses how these techniques can be applied to gender classification, emotion detection, channel identification. Although highly technical in nature, this presentation is framed in common, everyday terms that business decision-makers can understand.

Fraud countermeasures are one of the most relevant topics today. Many companies rely on voice biometrics to obtain natural and usable identity verification services. However, security breaches could be caused by recordings, audio manipulation, and even synthetic voices. Learn about latest development in biometric countermeasures systems and advanced spoofing attacks detection with the most recent deep learning training schemes.