INTERVIEW: Konstantin Simonchik, Chief Science Officer, ID R&D

INTERVIEW: Konstantin Simonchik, Chief Technology Officer, ID R&DFindBiometrics President Peter O’Neill recently spoke with Konstantin Simonchik, Chief Science Officer, ID R&D. The interview begins with some background on the biometrics company, which focuses largely on voice, speech, and behavioral recognition technology. The conversation moves to the various drivers causing the current boom in voice and speech recognition, and Simonchik further elaborates on how these technologies in particular are enabling the frictionless user experience that is in such high demand. The discussion turns to the importance of liveness detection in today’s biometrics landscape, gives way to talk of ID R&D’s vertical market focus and a novel case study, and concludes with a preview of what to expect from the company at next week’s Money20/20 conference in Las Vegas.

Read our full interview with Konstantin Simonchik, Chief Science Officer, ID R&D:

Peter O’Neill, President, FindBiometrics: Can you give our readers a bit on the background of ID R&D for those who are not aware of your past efforts?

Konstantin Simonchik, Chief Science Officer, ID R&D: ID R&D is a company that has been around for a couple of years now. At the core of our company is a very strong R&D team and that was the idea at the onset – to build a very strong R&D team with very strong R&D capabilities that would be able to bring new biometric capabilities through the investment of the core technologies. So, to make new user experiences and new levels of security possible through the application of the latest scientific breakthroughs and artificial intelligence and the methodologies of deep neural networks and convolutional neural networks. That is where the company is coming from, that is the core idea: to bring the latest scientific breakthroughs and apply them in the biometric field to enable new user experiences with new levels of security. They will bring this to the market specifically in voice biometrics, in liveness, in behavioral biometrics, and apply them to cases such as remote login on web and mobile platforms, and in conversational interfaces.

FindBiometrics: Over the past several years we’ve seen the voice and speech recognition industry really grow. What is driving it right now? Why has it become such a popular modality in the last couple of years?

ID R&D: I think we’ve seen some type of seismic shift in the perception of voice in the last year or two because previously the application of voice biometrics technology was related to call centers and phone channels. In the last year or so we have seen voice emerging as a new interface. Basically, if you look at the history of the technology and the way people interact with technological capabilities and consume them, the first wave of that came with the web boom, so we have been doing all of our transactions on the web. Then along came the smart phones and the ability to do transactions and services on the app. Now we are seeing a third phase and the use of a conversational interface. A good example here is Alexa, and the amazing success of it, and that more companies are building similar devices.

You can have the interface based on the free dialog with the consumer. Now we are seeing that the same principles are being applied to mobile platforms and a greater number are available through simply talking into your device or IT platform such as cars, connected home, connected devices, etc. So, voice essentially got out of the cocoon of the call center, and went into the wilds of mobile and IT world, and became this third interface that enterprises can communicate with their customers on. And of course, what came with it was the need to authenticate users using the interface in a secure and frictionless way, and that is what is really driving the growth in voice biometrics right now.

FindBiometrics: You mentioned frictionless which is a very hot topic these days. Can you elaborate more on that particular aspect of voice technology?

ID R&D: Absolutely. Frictionless is one of the key concepts now because the overall acceptance of biometrics is constantly increasing and the experiences that these technologies can bring to the table are getting more and more frictionless. We are asking the user to do less and less to authenticate when consuming personal services and conducting transactions. The same is going on with voice. The technological abilities that are available allows for a frictionless experience.

When we talk about a frictionless experience, we talk about the ability to use the free speech to communicate freely with the technology and have a very high level of security. And that is achieved two ways. One is the investment of the voice biometric algorithm, and what we have seen for instance at ID R&D is that earlier this year we introduced a new generation of our voice biometric solution based on Convolutional Neural Networks and x-vectors. What this new generation of voice biometrics allows is an extremely and highly accurate biometric authentication of free speech scenario where we can see ERR go as low as 0.3 percent, which is extremely low compared to what this technology was able to achieve before. But more importantly less and less speech is required for the purposes of verification where we can provide the customer the ability to do that identification based on just a couple of seconds of speech, versus six months ago when the technology needed 7-10 seconds to complete verification. So, we are seeing huge leaps forward in the ability of the voice biometrics.

The second important part of this equation is anti-spoofing, or the ability to distinguish between the user and the recording and the user and a computer-generated voice, and to do it in a frictionless way without actually asking end users to do anything specific for this test. We released new algorithms that can successfully distinguish between a live user and a recording or a live user and a computer-generated voice. When we look at the biometrics right now the technology can provide this comprehensive security that is based not only on the correct matching of voices but also on the ability to protect this match against various types of spoofing attacks. And we can do all that in a free speech scenario in the test independently.

FB: That is a lot of information. Liveness, as you mentioned, is the other hot topic out there right now. It is amazing to see how the industry has progressed in the last six months. You also covered a lot of vertical markets that are new to the industry – IoT, autonomous vehicles, robotics – what other vertical markets are you concentrating on?

ID R&D: We are seeing a lot of opportunities in a number of verticals. Of course, financial institutions remain a strong vertical for biometrics. Also, any type of connected devices and the type of connected devices that can be used for this conversational interface experience is constantly growing. We are working with a number of car manufacturers, large telcos, and large technology companies that provide the type of voice enabled capabilities.

One of the very interesting use cases that we are working on is a protection of the so-called ‘wakeup word’ from unintended activation using the core anti-spoofing algorithms. Essentially what we are talking about here is that a lot of the voice-based systems rely on the ‘wakeup word’ to start the conversation. So, you say “Hey Alexa do this” or “Hey Alexa do that” etc., and the problem is that the technology is constantly listening to the human and waiting for the ‘wakeup word’ to start performing its services and it can’t actually tell that the voice actually comes from a person in the room vs that voice on the TV or radio or through a loud speaker – which leads to a lot of of unintended activation of the voice systems with very unfortunate consequences related to that. So, what we can see right now is that we can do the anti-spoofing capabilities and protect the voice interface from not only fraudulent attempts but also from situations like that.

FB: You mentioned that financial services is one of the areas that you are focusing on; will we see you at Money20/20 in October where people can potentially come and see your products first hand?

ID R&D: We are actively working with a number of financial institutions here in North America, in Europe and Pacific Asia and there is a very strong interest in the voice conversational interface simply because the user or customers that are relying on voice in some other capacities are demanding the same type of service from their financial institutions. So, we are seeing tremendous potential there and of course we will be at Money20/20 so stop by our kiosk.

FB: Well Konstantin, as always, it is so nice to hear from one of the industry leaders bringing us up to date on all the tremendous advances in voice and speech recognition. Thank you very much and I look forward to seeing you at Money20/20 in one week!

ID R&D: Thank you Peter, I will see you there!