Show and Tell : Building Secure Voice Biometrics with Ryan Roberts

Voice Biometrics promises to enable a host of new, security-critical use cases on mobile and hands-free devices. From verifying users on mobile phones to unlocking your front door, voice commands will be analyzed to automatically authenticate the speaker and authorize command execution.

Voice biometrics adoption is experiencing strong growth, driven by the need for foolproof security methods for technology-driven systems and the popularity of voice assistants such as Alexa, Google Assistant, and Siri. But how do we gain confidence in the robustness of the underlying Voice Biometrics technology in the presence of different acoustic environments, noise conditions, and a host of other factors?

A single false identification could enable an attacker to compromise a user’s voice command security. In our second Show and tell event, we will be giving the stage to Ryan Roberts, who will provide insight into the approach used to systematically train and test this voice biometric AI technology.

Show and tell event series by Ai Kenya is meant to provide a platform for guest speakers to showcase their projects and consequently get feedback and collaborations from event attendees.

The Ai Kenya show and tell series focuses on projects and research efforts in the field of machine learning, data science and robotics.

Show and tell event objectives include:
> To showcase projects that community members or guest speakers are working on in machine learning, data science or robotics.
> To enable speakers to get feedback and possible collaborators on their projects
> To share insights or key points in specified projects or topics being presented
> To spark interest and encourage research in the projects being showcased or developed

In this show and tell event, attendees will discover:
>Overview of a Voice Biometrics algorithm.

>Importance of a data-first and test-first approach when developing high quality AI algorithms.

>Data augmentation techniques for increasing the test set.

>How to automatically and systematically test a voice biometric algorithm’s performance in a Continuous Integration environment.

Finally, Ryan will lead a code walkthrough/demo demonstrating how to characterize and visualize key performance characteristics of a binary classifier, before giving a demonstration of the Voice Biometrics technology in action.

About the speaker
Ryan Roberts has been working in the consumer electronics embedded software field for over 10 years. Based in London, Ryan currently works for Cirrus Logic( https://www.cirrus.com ), a semiconductor company that designs audio hardware and software for mobile devices, where he leads a software team to create a secure, embedded voice biometrics product targeting mobile phones.

Event Program
9.30am - 10am: Attendees check in
10.00 - 10.10 am: Introduction keynote by Alfred Ongere
10.10am - 11.30am: Overview of Voice Biometrics and demo by Ryan Roberts
11.30am - 11.50am : Q n A
11.50am - 12.30 pm: Networking, event ends, attendees leave at their convenience

Kindly do not book a ticket if you are not sure of attending the event.