Voice Biometrics

Voice biometrics, like face biometrics, are natural to use. When we cannot see someone but can hear him/her, that can be sufficient for us to recognize who it is. This biometric's greatest strengths are its size and noninvasiveness, as seen in Figure 9-3. The greatest weaknesses come from the FAR and FRR. What follows is a breakdown of each characteristic for voice biometrics.

Figure 9-3. Scores for voice biometrics.

Acceptance: 8.5

Voice biometrics rely on the user 's speaking for recognition, which is obviously something that users do all the time. The voice biometric has wide acceptance. What prevents it from having the same score as, or a higher score than, fingerprint biometrics is the fact that it is passive.

Easy: 5

The ease of use of voice biometrics needs to improve. The required time to train a system and enroll, coupled with the need for clear speech and low ambient noise, can make voice biometrics difficult to use. For this reason, it did not score well in this category.

ROI: 5.5

The cost of the hardware to use with voice biometrics is relatively low. A decent-quality microphone is relatively affordable. The cost of deploying the microphone is normally low, as most computers have sound cards with a microphone jack. What diminishes the ROI is the user's need to use the biometric daily. Voice biometrics can take a long time to train and enroll. They are also very susceptible to background noise and changes in the user's voice. For these reasons, many calls to the help desk could be generated, thus lowering the overall ROI.

Deployable: 8

A voice biometric for network security is relatively easy to deploy. The microphone needs only to be plugged into the sound card of the user's machine. That is relatively quick and easy to do. Once the microphone is deployed, it needs to be tested and possibly re-adjusted to minimize background noise.

Since additional time is needed for testing and placement of the microphone, voice biometrics scored slightly lower than finger biometrics.

Noninvasive : 9

Since the voice biometric is natural, users find it noninvasive. We are all comfortable with using our voice for recognition. A perfect score was not given since some may feel that voice biometrics can be used as a passive tracking system. For this reason, it does not have the 100% acceptance that the ideal biometric would have. One way to increase acceptance to almost 100% is through the use of a privacy-positive biometric policy. This way, the users would feel more comfortable with the technology's use.

Mature: 7

The use of voice identification has been around a long time. We are accustomed to using the sound of other peoples' voices to identify them. However, the use by machines of voice biometrics is relatively new and still needs additional research and time for it to mature.

FAR: 6

As discussed in Chapter 7, for any biometric, you must decide on the tradeoff between security and user convenience based on the algorithm chosen . The voice biometric is also fairly susceptible to spoofing attacks.

For the above reasons, voice biometrics scored low on the FAR measurement.

FRR: 5.5

For voice biometrics, there can be a very high FRR. This is due to the nature of the biometric. The voice changes over time, and can also vary from hour to hour depending on physical health and the ambient environment. Ambient noise can cause the FRR to increase. The enrollment procedure can also lead to an increased FRR. If the enroller and user were not diligent enough in their training of the voice system, the user could have a very hard time authenticating.

Size: 9.9

The size of the microphone is very small and will use almost minimal real estate on the desktop. It did not score as high as the ideal biometric, however, since something still needed to be deployed.

Habituation: 7.5

The use of voice biometrics can require the user to be highly habituated. The user needs to present his/her voice at the right pitch, tempo, and cadence. These things can be difficult for people to do, so the time needed to be come habituated to the degree required could be relatively long.