Aculab voice biometrics system authenticates non-verbal speakers
Voice biometric expert Aculab has recently published a blog post illustrating the accessibility features of its VoiSentry voice recognition system.
Written by digital analyst Reece Wheeler, the post describes the importance of developing voice biometrics solutions designed to support non-verbal speech.
According to Wheeler, while voice biometric solutions can be more intuitive and secure than traditional passwords and PINs, they must be developed with accessibility features in mind, lest they end up acting as a barrier for people with speech-related disabilities.
“A good language-independent Voice Biometric system relies on the voice characteristics rather than the actual words spoken,” Wheeler explained. “This means it can still be used by people with impaired speech as long as they can vocalize to some extent.”
The VoiSentry system is reportedly built on these principles, the digital analyst explained, and is capable of extracting many diverse vocal traits by using a hybrid approach.
Specifically, the technology uses a mix of artificial neural networks, signal processing technology, and analytic linguistics, to create an authentication process as frictionless as possible.
Wheeler, who has cerebral palsy and limited speech, tested the VoiSentry demo software to assess its voice recognition capabilities.
“Honestly, I was impressed to find that the application was able to identify and verify me every single time,” the digital analyst said.
“I made numerous attempts, making different noises to challenge the software. In the end, it demonstrated that it was able to pick up the nuances of my voice in order to get a verification,” Wheeler added.
Aculab’s platform has also been confirmed as secure, with the company being recommended for ISO 27001 accreditation last January.