There are a number of companies that have or will soon be releasing new services for integration with voice biometric services. So what’s possible with voice biometrics?
Registered user authentication. The ability to determine whether a given user is speaking and then unlock access. This is the case with “OK Google” or “Alexa” registered wake words. With this, users can allow their device to be unlocked only when they speak. Registering usually involves repeating the wake word a few times.
User identification. This is when you try to identify a user against a previously acquired sample of that person’s voice. This ID can be pulled from a list of several individuals to potentially millions, in the case of government surveillance software.
Characteristic identification. Can specific traits of an individual be extracted from a voice sample? Accent, gender, age, and even emotion might be characteristics that are extractable.
In addition, these authentications can be done on the device and in real time or they can done off device and non-realtime. The embedded versions can also run “on-host”, like any other application, or “on-target”, as a low level, low power service.
As the cost to license and implement these services come down, we can expect very interesting interactions to be built.