Siri must be running on nearly 1-billion devices worldwide; Mac, iPhone, iPad, Watch. Add ’em all up and Siri may well be the most visible and most used of the speech recognition systems in our gadgets.
The problem with Siri can be summed up in two considerations. First, speech recognition. Second, actions. Of the first, Siri’s ability to understand what we speak has improved, and the new Siri in iOS 11 even sounds far more human. As to the second, Siri– as well as Google Assistant, Microsoft Cortana, Amazon’s Alexa– can perform some specific parlor tricks, a few of which differentiate each from the other, but not enough for anyone to call any of them a truly personal or intelligent assistant.
Microsoft’s engineers claim their own home-grown speech-recognition system can match professional human transcribers; those who listen and transcribe what they heard. The error rate among humans is 5.1-percent (remember, these are professional transcribers, not your spouse or children or anyone else you know) and Microsoft finally got there. The test itself is a database of 2,400 two-side telephone conversations between people who don’t know each other but have U.S.-based accents.
In other words, these speech recognition systems are pretty good at recognizing what they here with a degree of accuracy similar to humans.
Anybody see a problem there? Humans are notorious for not fully understanding or comprehending what each of us say to the other, so a device and technology that purports to transcribe what is said and turn it into useful functions or actions seems like it isn’t just around the corner, but perhaps not even on the same planet.
Yes, I use Siri. I like Siri. Siri does a better job transcribing text than listening to and understanding commands and requests for action.
Pretty much like a human being, right? Maybe creating a machine that behaves like a human isn’t as difficult as once thought. For what it’s worth, Siri has improved over the years, and Apple has doubled down on the technology, even more evident with iOS 11 which will end up on about 1-billion devices by this time next year.
The bar is being moved forward. 3D Face ID will show up in the next premium iPhone, and likely in the HomePod speaker system later in the year, and certainly in future iPhone, iPad, and Mac models over the next year.
That’s when it gets interesting– the marriage of 3D Face ID technology with Siri. I envision a few conversations like this:
Siri: “Jeffrey, you look sad. Is everything OK?”
Me: “No, Siri. We’re all out of whiskey.”
Siri: “Shall I have some whiskey delivered to you.”
Me: “I love you, Siri.”
It could happen.