I received my Amazon Echo recently. I ordered it merely as a curiosity and to generally stay aware of industry trends. But after just a few days of using it at home; I love it enough to prompt dusting off this blog after almost two years of no posts!
The first thing that caught my attention was the sheer accuracy of its voice recognition. The state of the art is already pretty good; the dual pillars of cheap, persistent computing power (i.e., “the cloud”) and renewed interest in machine learning brought us Siri (and their Microsoft & Google equivalents). In my experience, these have been accurate more than 95% of the time, a long way ahead from the days of offline speech recognition (à la Dragon Naturally Speaking). The bar is already high, but I’m comfortable saying Amazon Echo’s voice recognition is definitely better than Siri’s or Google’s.
I’m not sure how they pull it off. Maybe it’s not because of better software, but simply better hardware. The Echo has an array of 7 microphones that are always listening and wake up as soon as you say “Alexa”…
There is an element of genius in packaging it as a standalone cylinder that sits in a somewhat central location in your home. The experience of being able to talk to it, hands-free, from a wide range of places in your home sounds like it wouldn’t be that big a deal… until you actually do it. Then it makes having to find your phone, pick it up, and push a button to make it do something seem archaic and boring.
Now, the only problem is that even though Echo understands what I’m saying, it doesn’t always know how to respond. The companion app often shows an accurate transcription of what I said, but since it did not fall into one of the categories it’s designed to handle at the moment, I don’t get a satisfactory response.
But there are relatively easy ways to fix that. The Echo API is currently invite-only, like the ability to purchase the device itself. As more developers get their hands on it, we’ll start to see many interesting things happen. Some developers already hacked their way into controlling their home lighting (Amazon now officially supports integrations with Philips Hue). I’m itching to make it control my home theater system, currently riddled with half a dozen remotes.
The idea of having a single hub in your home that’s always listening and can control all your other devices is incredibly appealing to me. For one, it means that your TV, Xbox, Nest, and other home devices don’t have to build in (often bad) voice recognition systems themselves.
Finally, I was impressed by how handily the Echo passed Larry Page’s “toothbrush test”. Even with its somewhat limited functionality, my wife and I have already used it several times every day since we received it — for things ranging from compiling a shared grocery list, adding reminders, setting alarms, and playing music.
I think Amazon is onto something big. Priced at an aggressive $99, the Echo has the potential to make it into the living room of a large majority of households in the US (and the world, eventually). Will we now see a flurry of competitors from tech giants and startups?