Leading Technology

The strength of ID R&D is our technology team.  Our focus is research and development in the science of biometrics; that’s why R&D is in our name.  Learn more about our core technology for voice biometrics, voice anti-spoofing, face anti-spoofing, behavioral biometrics, and acoustic event detection on this page.  These technology components, all available as SDKs on multiple platforms such as iOS, Android, Linux, and Windows, form the basis for authentication applications visible on our Products and Services page.  

Problem Solvers

We started ID R&D with the vision to use technology to improve the onerous task of authentication.  Our view of using technology to solve problems extends beyond authentication, however.  Use our expertise with audio, speech, and artificial intelligence to address your needs.  For example, responding to customer need was the driver for our Acoustic Event Detection SDK, an SDK that uses Deep Neural Networks and signal processing analysis of sounds to detect non-speech events of interest.  In short, ID R&D is here to help you solve problems, not just sell you products.

Voice Biometrics

ID R&D’s voice biometrics capability starts with the science of biometrics.  The technology team applies its expertise in speech processing and artificial intelligence to create state-of-the-art software.  ID R&D’s latest voice biometric engine, IDVoice™ 2.0, fuses proprietary x-vector technology, deep neural networks (DNN), convolutional neural networks (CNN), and ID R&D’s patented p-vector technology to achieve low error rates for both text-independent and text-dependent use cases.  The result is exceptional performance: The Equal Error Rate for IDVoice™ 2.0 against an industry standard microphone text-independent (free-form speech) database is 0.30%. Text-dependent performance is 0.39% against the RSR2015 database.   

Features of IDVoice 2.0:

  • Use with any language
  • Text-dependent, text-independent, and prompted modes
  • Fast voice activity detection and speech endpoint detection
  • Industry-leading matching algorithms for enrollment, verification and identification
  • Accurate signal quality estimation including signal-to-noise ratio and net speech length estimation
  • Configurable SDK footprint – from 2MB on mobile platforms to highly accurate and faster server-based packages
  • Cross-channel compatibility using calibration setting, no retraining
  • FIDO-compatible SDK for mobile devices

Unlike other solutions, ID R&D’s core voice technology works in any language without retraining, works across channels with a calibration setting, and is designed from the beginning to be noise-tolerant.  You can use ID R&D voice biometrics with near-field and far-field microphones.

Voice and Face Biometric Anti-Spoofing

Features of voice IDAnti-Spoof™:

  • Works in any language
  • Works even with short speech segments – from 1 sec
  • Configurable SDK footprint – from 3MB for mobile platforms
  • FIDO-compatible SDK for mobile devices


Fraudsters constantly attack security systems to find weaknesses.  Biometric systems are no different. “Spoofing” refers to an attack on a biometric system that attempts to use fake biometrics of the valid person.  For voice, the types of attacks are synthetic speech, voice conversion or imitation, and recording replay.

ID R&D invests in anti-spoofing technology to determine if a given audio recording contains a genuine voice or a spoof.   ID R&D applies multiple techniques, including specific speech feature extraction, Gaussian Mixture Models, factor analysis, and deep convolutional neural networks.

For facial biometrics, the simplest spoof attack is a photograph.  A typical anti-spoof strategy is to require the user to move their eyes or look in a different direction.  ID R&D technology is different. The ID R&D technology team developed proprietary algorithms to detect facial liveness without asking the user to do anything, enabling a zero effort authentication experience while preventing a spoof attack.

ID R&D’s Voice Anti-spoofing technology is one of the leaders in the ASVSpoof 2017 challenge (http://www.asvspoof.org/).

Behavioral Biometrics

In contrast to voice and face biometric technologies that make use surface characteristics of external human physiology, behavioral biometrics use patterns relating to human activity.  Surprisingly, these patterns are consistently unique for each person to the point of being useful as an additional biometric factor. ID R&D’s focus with behavioral biometrics are human activity patterns we can measure on widely used devices — smartphones, PCs, tablets, and other common devices with human to machine input.

Our IDBehave™ technology, the top performing technology in the 2017 Keystroke Biometrics Ongoing Competition, encompasses behavioral biometrics such as keystroke dynamics, accelerometer and gyroscope data, mouse movements, and any other data available from the device.   Specifically, IDBehave™ uses the manner and rhythm in which an individual types characters on a keyboard or keypad and synchronizes it with other sensor data. From these patterns, we develop a unique biometric template of the user’s typing pattern or of using a screen on the device.  For future authentication, we check the behavior against the template.

We are able to capture raw keystroke measurements from almost any keyboard, including through web browsers.  The most basic statistics we record are the time a key is pressed and the time between when the key is released and the next key is pressed.  We then process the keystroke timing data and other sensor data through a unique neural network to determine a template for future comparison.  

ID R&D participation in Keystroke Biometrics Ongoing Competition (KBOC) organized by IEEE documents ID R&D’s leadership position in the Keystroke Dynamics segment (2017) (https://sites.google.com/site/btas16kboc/home).

ID R&D Behavioral Biometrics

The features IDBehave™ provides are:

  • Small SDK footprint as low as 1MB
  • Six symbols of biometric pattern or more
  • Bot detection (mobile, desktop, server platforms)
  • FIDO compatibility SDK for the mobile platform

The primary spoofing attack on keystrokes is a bot.  IDBehave™ recognizes bot attacks because human behavior is difficult to mimic through a computer program over time.

Acoustic Event Detection

In addition to verifying and identifying voice, ID R&D applies its Artificial Intelligence expertise to other types of audio.  Non-speech sounds, called “acoustic events,” fall into this category.

Acoustic events, such as hearing a baby crying or an elderly person experiencing a coughing attack, are often cause for action.   However, manual monitoring, either in close proximity or remotely through a baby monitor device, requires attention and to be in hearing distance.   This is not always possible.  Imagine an elderly care facility with many patient beds and a nurse at a central station trying to listen to monitors for all rooms.

The solution is ID R&D’s technology to automate the listening for these events and alert an application programmatically that the event occurred so in turn a human may take the appropriate action.

As with voice verification and identification, two types of errors may occur:  A false positive and a false negative.  Of these two errors, false positives, that is, thinking an event occurred like a coughing attack when someone is just shuffling paper near a microphone, present an annoyance to the person monitoring.  If numerous false events appear, the monitoring person is inclined to simply shut off the system.  Therefore, ID R&D targets false positives to be less than 0.1% as even 1% false positives over a large number of possible events creates annoyance.

Acoustic event detection enables a new range of possible applications, especially as IOT devices with small microphones become ubiquitous.  Contact us if you have a need for an event not already on our list.

Acoustic Events Biometric Detection

ID R&D’s Acoustic Event Detector listens for the following sounds:

  • Baby crying
  • “Help”
  • Coughing
  • Moan
  • Scream
  • Dog barking