Voice as Your Verifier: Introducing Voice Verification for PingOne Verify 

In a world powered by digital experiences, it’s hard to know who is the anonymous person on the other end of a phone, or even if it is a human (more on that point later). Voice verification, also known as voice authentication, uses the unique characteristics of an individual's voice to verify their identity. It is a biometric authentication method that analyzes various voice features such as pitch, tone, and pronunciation to determine whether the speaker is authentic. It can be a completely passive form of authentication and eliminate friction for the user.

The user’s voice must be enrolled for Voice Authentication. This may be done by having the user record 3-5 samples during a self-service enrollment. Voice enrollment may also be done passively by harvesting audio files from an IVR system or collecting a voice sample inside an engagement with a call center agent.

To verify an enrolled user’s identity, the biometric voice recognition system captures a new speech sample in 1 of 3 ways:

1. The user’s voice is captured through the Interactive Voice Response (IVR) system before reaching a live agent. This is considered text-dependent voice verification, where a person speaks a specific passphrase, usually consisting of a few words. Remember “My voice is my passport; verify me”? Text-dependent voice verification is not as secure as other methods; more on that later.

    

    

   

    

2. Next, there is text-independent voice verification, where users are passively authenticated during an engagement with a live agent. The user can say anything, enabling authentication to quickly happen in the background during their normal interaction with an agent, IVR, or application. This is a more secure method for implementing voice verification as it reduces the vector for spoofing with a synthetic voice.

    

3. Finally, users can be sent a link via email or SMS to perform the verification on a web-based or mobile application verification flow. This provides an additional level of security in that it simultaneously proves device possession. Again, liveness checks are supported through this channel.

An important benefit of authenticating a user digitally is collecting a sample with the superior audio quality that offers increased security capabilities for verification. Mobile devices support a 16kHz bandwidth as opposed to a landline which is only 8kHz. PingOne Verify performs passive liveness detection to stop recorded playback attacks exclusively over a 16kHz channel. Today, no voice verification service available in the market can perform liveness detection using 8kHz frequency. The audio quality just isn’t good enough.
 

Once the audio sample is collected:

• PingOne Verify processes it and extracts a template. 

• The new template is compared against the enrollment template.  

• A strong match between templates indicates that the same person spoke both samples, thus verifying the person’s identity. 

• PingOne Verify returns the probability and confidence score of the comparison in its output.

• You can optionally enhance the master voiceprint with each successfully verified voice template for stronger voice modeling.  

Call Centers: Voice verification can be integrated into Call Centers for any business to provide user authentication or transaction authorization. Users can verify their identity by naturally engaging with a live agent or IVR system. Voice verification should be used as part of a layered approach to user authentication. 

Phone Banking: Banks can use voice verification to authenticate customers over the phone. Customers can securely access their accounts, make transactions, and perform other banking activities using their voice as a unique identifier.

Major benefits

• Much stronger than Knowledge-based Authentication, the traditional way to verify users over the phone. Users put far too much information online, making it easy to find out what high school a person attended or their mother's maiden name. 

• Free agent time by eliminating cumbersome KBA Q&A.

• Secure the customer experience with fast, frictionless authentication. Coupled with device possession or SIM checks, voice biometrics is a highly secure way to authenticate users in a call center.

Is it safe? 

Voice verification is both safe, and highly effective. As humans, we use more than 70 muscles to create speech, resulting in something unique to each of the 7.8 billion people on the planet. With regards to the implementation of voice verification in PingOne Verify, the AI does not recognize the words being used, instead voice biometrics recognize the unique characteristics of the user’s speech. 

Will voice verification still work if the user has a cold?

Absolutely! A strong voice recognition biometrics algorithm will continue to work as expected and with high accuracy even if the user has a cold. However, if the voice sounds entirely different, then by design, the person would not be recognized as a match with the enrollee. In extreme cases like this, the user should be required to use a different form of authentication.

Every form of biographic verification will invariably find the edge case. As we’ve seen with fingerprint, users with callus-buildup, such as from manual labor, are unable to use it.