Voice recognition
Dear HIW, I recently purchased a Google Home speaker. I was wondering how voice assistants can use voice activation (‘OK, Google’, ‘Hey, Siri’ etc.) and differentiate voices (how it can tell my voice apart from my dad’s). How long ago did these sorts of mechanisms come into everyday use? Caden Blythman Voice assistants in our gadgets work by listening out for their activation phrase, such as ‘OK, Google’. To do this, they are always linked to your device’s microphone, but they will only start ‘paying attention’, as it were, once you say its specific key words. The feature that recognises different voices in Google Home devices is called Voice Match. It can recognise up to six different voices using technology that links each speaker to their preferences, such as their favourite news source or the expected commute time to their place of work. When setting up Voice Match, users repeat the phrases ‘Okay, Google’ and ‘Hey, Google’ twice. This trains the Google Home to recognise the tone and pitch of your voice by creating a voice profile that it matches every time you speak to it. Other smart home assistants are introducing this system, including Amazon’s Alexa. Voice recognition software itself has existed since 1961, when IBM introduced their Shoebox tool, which could recognise just 16 words: the digits 0–9 and arithmetic commands such as ‘plus’ and ‘minus’ – effectively a basic voiceactivated calculator. Clearly the technology has evolved dramatically since then, with smart speakers able to understand and respond to normal speech with the help of rudimentary artificial intelligence software. These technologies have become more widespread in recent years ever since Apple introduced Siri in 2011 and Amazon launched their first Echo speakers in 2014.