An assistant is so much more useful when it knows you
Earlier this year Google rolled out voice profiles for its Google Home assistant, and Amazon has recently added similar functionality to its Echo range of devices.
What sets voice profiles apart from regular voice recognition is the ability for your device to determine who’s speaking, and therefore respond in a more personalised manner.
When you speak to someone you know in real life, what you say and how you say it will be guided in some way based on who that person is. If you don’t know who you’re talking to you’ll have to go through several steps to learn something about them: name, age, interests, etc.
With voice profiles, a host of information about each speaking person can be called upon in an instant – Google and Amazon accounts hold tomes of data relating to who you are.
So, what kind of stuff is possible once your AI assistant knows who you are?
- John asks Alexa to call mum and Alexa knows to call his mum and not his wife’s mum.
- John asks Alexa to play some recommended music, and he gets songs based on his love of rock music, not the nursery rhymes his kids listen to.
- Reminders and diary entries will be much more intuitive to set and retrieve.
In all these cases the user can speak more naturally, without having to identify who’s talking by speaking in the third person.
And then there’s online shopping. Neither Google Home or Amazon Echo support child profiles, but this would be a great way to prevent the little'uns in your life from ordering everything on their Christmas list.
There’d be no need to mess around with passwords, as the voice becomes the password. If little Timmy hasn’t got clearance to spend £100 on Peppa Pig toys, the assistant won’t let him.
And the power of knowing the speaker becomes even greater as more information is learned – something which is possible thanks to voice profiles.
If John always comes home and asks for certain lights to come on, music to play and the oven to be pre-heated, the virtual assistant of the near-future could learn this. These learnings could then be used to automate the actions when John says hello as he opens the front door.
Then there’s home security, which could benefit from recognising the voices of those who live in a given home. When you enter the home the assistant could ask, “who’s that”, at which point a response from a known family member would disable the alarm system. A unrecognised voice would require a password/phrase to prevent an alarm from sounding.
There are obvious flaws in using voice, such as changes in a user’s speech or the use of a recording, and these would certainly need to be addressed.
Slight concerns aside, this is a great example of how advancements in tech and the growing pot of data stored on each and every one of us can combine to offer new and exciting functionality.