Identifying a person from their voice
Your voice is unique. The shape of your vocal tract and the way you move your mouth when you speak produces distinct sound patterns. It is possible to measure these characteristics and create a voice print – a representation of the biometrics of your voice.
The term voice print can refer to a sample of recorded speech, or to the expression of that speech as a mathematical formula. This enables a person’s voice to be represented graphically, and thus to be used in voice identification systems to verify identity.
Since its commercial launch in the late 1980s, voice recognition technology has found its way into a variety of applications. It is easy to use – dependent only on a microphone and a software package – and also quick, versatile and contactless. One of its most widespread applications is in telephone banking, where a customer’s voice acts as the password to remotely access their account.
Voice prints have also been integral to the development of digital personal assistants. Back in 2015, Apple integrated voice identification to their personal assistant, Siri, enabling users to unlock their phones via voice command. New generation personal assistants such as Google Home and Amazon Alexa provide users with the option of registering their voice print. This enables different users to receive personalised responses from the assistant, as well as facilitating secure voice shopping and even donations to charity using voice commands.
Although voice printing is becoming more sophisticated, it is still possible to fool this technology. In 2017, a BBC reporter hacked into his non identical twin brother’s bank account by imitating his voice. The persistent vulnerabilities of voice identification technology mean that it is strongest when used in tandem with other security measures, such as pin codes.
For more insights into disruptive technologies subscribe to our free weekly newsletter.