Call Center Voice Verification

Solution enables to quickly and reliably confirm speaker identity upon voice when contacting call center.

Verification can be performed upon both Customer-Agent and Customer-IVR interaction.

Using biometrics provides higher safety level of remote verification in comparison with the standard Question-Answer approach. Safe verification makes it possible to expand the range of services provided by call centers and attract new customers.

The ability to quickly and easily receive a service or solve a problem remotely, without attending an office or a sales point, increases the convenience for customers and their loyalty.

Key Features

  • Safe and secure. Automatic verification is more reliable than human one*. Modern approaches based on deep neural networks (DNN) and CX-vectors underlie technology developed by STC-innovations.
  • Convenient and simple for customers. Enrollment of biometric samples and further verification are performed in the background on the basis of free speech. There is no need to pronounce any predefined phrases or answer additional questions.
  • Voice-independent. Solution can build voice references for speech in any language that enables to use it all over the world.`
  • Fast results. From 5 to 10 seconds of pure speech
    are enough to perform comparison.
  • Support for popular PBX, IVR and CRM systems.
    Integrative with Avaya, Genesys, Cisco, Asterisk, Noda.
  • Single enrollment for agents and IVR. Verification is
    performed on the basis of just one biometric reference.
    Enrollment is performed only once.
  • Flexible scalability. Installation is possible for call centers
    of any size: from several dozens to several thousands
    of agents.

* The Conversation Global, UK

General Scheme of Work

  1. CRM/IVR system sends verification request with customer identifier specified.
  2. Solution receives sound from the telephone line of a call center.
  3. Solution generates biometric voice template of the caller and compares it to the voice reference stored in the database.
  4. Results are returned to CRM/IVR system. Results contain percentage of similarity between newly-built biometric template and biometric reference stored in the database.

Minimum System Requirements

Biometric platform:

  • 16 Core CPU / 32 GB RAM / 100 GB HDD
  • CentOS / RHEL 7.6 x 64

Telephony integration server:

  • 4 Core CPU / 8 GB RAM / 50 GB HDD
  • MS Windows 2012 x 64