User Training for Voice Biometrics

Training for Active Systems

For active voice biometric systems, such as RandomPIN™ or Static Passphrase, the person speaking is knowingly (and actively) participating in the voice biometric enrollment or verification process. Thus, providing users with training on how to interact with a voice biometric system can have a significant impact on your system's results. VBG has years of practical experience with our clients and partners that proves this point time and time again.

When considering how to inform end users about your voice biometric process, you can and should be explicit with your instructions and guidance. Your goal is to encourage positive end user behavior that in turn will help the system to operate optimally. Proper training will help you to experience greater accuracy, achieve higher levels of user satisfaction, and also minimize support requests related to voice biometrics.

Below are several principles to keep in mind as you educate your end users:

• Remind users that the purpose of the system is to both save them time and provide stronger security for their account.
• Provide enough detail on the process so the user understands the need to first capture a good voiceprint, so that is can be later used to conveniently verify them in just a few seconds.
• Because the quality and clarity of audio are important, tell your users not to use speakerphones. Speakerphones are designed to pick up sounds all around a room and they also create echo. Both of these issues can impact voice biometrics negatively.
• To the extent possible, have your users use the same device for enrollment and verification. Significant compression and tranmission differences can occur as devices change, so consistency is key to having consistent performance.
• Keeping with the theme of consistency, instruct your users to speak naturally and with their normal, conversational pace. Voice biometrics examines numerous aspects of how you speak, including pace, so speaking consistently is beneficial.
• Although perhaps obvious, the user needs to speak in an environment that has low noise levels. Loud and persistent background noises will interfere with the voice biometric system's ability to process speech samples properly, potentially leading to a false rejection.
• Use all VBG response codes and error messages to help interactively guide your users. For example, our API response will respond with a 90400 response code if the speech sample has too much noise. Your application should trap for this code and prompt them to try again or move to a quieter location.
• By keeping these ideas in mind, you will increase the effectiveness and efficiency of your voice biometrics system, achieving greater accuracy, and higher user satisfaction and acceptance, while minimizing your support costs.

What About Passive Systems?

In passive voice biometric systems, there is no specific prompting, so there is no specific end-user training required about how or what to speak. A well-designed passive system should leave the user entirely unaware that voice biometrics are being used (beside any required system disclaimers). Using a technique such as NaturalSpeech™ will ensure that naturally-spoken responses are provided.

However, end users will still need to be mindful of using speakerphones, trying to speak in noisy areas, etc. Thus, it is critical for developers of passive systems to use all of the VBG Platform's audio analysis and feedback mechanisms to properly and dynamically coach end users while in the middle of a session.