Voice Biometrics for Fraud Prevention

The human voice is unique

No two voices in the world are one hundred percent identical. They vary according to physical characteristics such as the size and shape of the larynx and the length and frequency of the vocal cord.

In addition, acquired characteristics such as dialect, accent, speech disorders or speech habits play an important role. This uniqueness is used by SpeakerVerification to perform secure and convenient verification, either alone or in combination with passwords, passphrases or a PIN.

Speech is perfectly suited for biometric identification: on the one hand, there is a high variability between different speakers, but on the other hand, there is a high consistency between different speech samples of the same person.

During the short login process, SpeakerVerification generates a unique voiceprint of the caller. When this user calls again later, the solution compares the voice with the stored voiceprint. If they are identical, the caller is granted access to the system. Otherwise, it is rejected and can be forwarded to an agent.

If required, SpeakerVerification verifies the caller not only at the beginning, but continuously throughout the dialog, for example, to detect attempts at fraud with edited voice sequences. This further increases security.

Unlike other biometric technologies such as iris scan or fingerprint, CreaLog SpeakerVerification can be used at any location where a mobile or fixed net telephone is available. So there is no need for the caller or customer to purchase new and unfamiliar end devices or install additional apps on the smartphone.