Combat Voice Spoofing Attacks

with IDLive™ Voice

Fraudsters constantly attack security systems to find weaknesses. Biometric systems are no exception.

“Spoofing” refers to a presentation attack using fake biometrics for a valid person. For voice, fraudsters use advanced speech synthesis, voice conversion or imitation, and recorded replay to try to spoof systems. IDLive Voice combats this with advanced voice liveness detection.

Voice verification answers the question, “Are you the person you claim to be?” Voice anti-spoofing answers the question, “Are you a real person?”. Liveness improves security when authenticating a known user and enables better, simpler identity proofing during new customer enrollment.

IDLive Voice capabilities include:
  • null
    Replay attack detection
  • null
    Computer generated voice detection (TTS)
  • null
    Computer modified voice detection (voice conversion)
Graph showing voice Anti-Spoofing using artifacts

How does voice anti-spoofing work?

Speech conversion, replay attacks, and TTS create certain signal artifacts that are sometimes indistinguishable by a human ear. ID R&D algorithms find and identify such artifacts to accurately determine liveness.

ID R&D invests heavily in voice anti-spoofing technology as we believe it is a critical component of strong authentication. Our IDLive Voice product leverages multiple techniques to passively determine liveness, including specific speech feature extraction, Gaussian Mixture Models, factor analysis, and deep convolutional neural networks that have been trained for spoofing detection.

Industry-leading anti-spoofing capabilities

ID R&D was the clear leader in the ASVspoof 2019: Automatic Speaker Verification Spoofing and Countermeasures Challenge for Logical Access (LA). The challenge is the largest of its kind and was designed to test the ability to distinguish between human voice and a synthesized human voice.

ID R&D was also a leader in the prior ASVspoof Challenge, ASVspoof 2017, which was designed to test biometric technology capabilities for “replay attack” detection (live voice vs a recording).

ASVspoof 2019 challenge results

Getting started with IDLive Voice

IDVoice Live is delivered as an SDK on multiple platforms including iOS, Android, Linux, and Windows, and as a Docker image. Our SDKs can be combined to deliver even more powerful capabilities.

IDVoice Verified is an ID R&D solution that combines IDVoice Live with the text dependent and text independent voice verification capabilities of IDVoice.

Want to learn more?

Menu