FFIEC Guidance: Authentication in an Internet Banking Environment (2001)
FFIEC Guidance: Authentication in an Internet Banking Environment Supplement (2005)
Have questions? Need more information?
Escalating incidents of identity theft, fraud and social engineering attacks continue to compromise existing data security measures. Traditional single-factor authentication approaches including passwords and challenge questions no longer provide the necessary safeguards for secure remote services. Biometric speaker verification technology uses the power of voice to provide the critical component in an effective multi-factor authentication solution and is the most comprehensive natural-language speaker verification system available.
Nuance FreeSpeech is a text-independent biometric speaker verification system that verifies a caller's identity during the course of a natural conversation. Content, language and accent independent, it transparently retrieves the biometric voice characteristics required for verification within seconds, eliminating the need for cumbersome authentication questions. Moreover, it enables contact centers to monitor all calls and detect fraudulent activity by employing a watch list mechanism. For more information, contact salesvoicebiometrics@nuance.com. |
Once a customer is transferred to a contact center agent and starts to speak, FreeSpeech listens in to the conversation and retrieves the speaker's audio for processing. The acquired audio is compared to the speaker’s voiceprint stored in the system, and a verification result is generated within seconds. The verification result is then transferred to the CTI, agent workstation or CRM system. As the conversation proceeds and more audio is acquired, further verification results can be generated.
Enrollment is performed while the user is engaged in a normal conversation with a contact center agent. The system records the first one or two conversations with the user, extracts the voice features that are unique to the specific person, automatically creates a voiceprint and stores it in a secure directory.
Verification is performed in real time in the background of a natural conversation. Captured samples are compared to the speaker's stored voiceprint and a verification score and decision are available within 5-10 seconds.
