insights and news

The Potential of Voice Recognition and AI System

18 June 2020

The Start of Voice Recognition

Capturing speech and transcribing it to text is the basis of the technology that is revolutionising the way people live and work. Over the last 50 years there has been exponential growth in voice technology. In 1952 the potential for voice recognition was first demonstrated. Then by 1976 computers could only understand around 1000 words. By the early 80’s that number was around 20,000, with continuous development until 1996 when IBM released the first commercial product, that could recognise continuous speech. 


In the last decade the development of Artificial intelligence technology (AI) has speared ahead the technology to unprecedented potential and turned it into a multi-billion Dollar market, expected to be worth $31.8 by 2025. AI is a term used to describe a series of related technologies that use large volumes of data to make predictions. So, in this case analysing the text that has been captured by voice recognition to predict the meaning and find a suitable response.


The machines ability to learn over time and become ‘better’ at predicting what is being said and its meaning has allowed the uses for voice recognition to grow, and the development of products such as Amazon’s echo and Apple’s Siri have quickly gained acceptance by consumers fr use of voice recognition and AI incorporated products. Voice recognition technology is today used by consumers to transcribe speech to text, set up reminders, internet searches, traffic and weather reports and get responses to simple questions and request.


The Use of Voice Recognition in Security Solutions

Voice recognition has also found a place in the security sector. The human voice is considered to be a type of biometric, known as a voiceprint. Each person has voice characteristic that are unique to them and can be used for the purpose of identification. Using voiceprint in this way is actually the easiest biometric to implement as it does not require the use of specialised equipment. With the continuous ‘learning’ potential of AI systems it means that the technology will become increasingly reliable and secure.


Voiceprint is already being used to authenticate customers on the telephone by banks. With its ease to implement, voiceprint can also create an added layer of security to existing multi-biometric identification processes.


Research has shown that voice recognition is favoured as a form of identification as it’s a non-contact and non-intrusive and easy to do. Although there is great promise, there are still potential challenges to Voice. The potential high error rate, due to variability in a person’s voice due to illness, mood changes, poor quality voice samples due to the presences of back ground noise are some issues yet to be overcome.