So I saw an ad on Facebook for this new Kickstarter. It’s called Car Wink. The concept’s pretty cool. It’s basically a tiny LED screen to hang on the back of your car. The idea is to facilitate communication with other motorists. It’s got various emojis you can display, along with words. The device is controlled with your voice, so you can use it while driving.
The use case is pretty clever. You’re on the road and construction up ahead is blocking your lane. A nice driver behind you gives you a space and lets you merge. You want to thank the driver, so you give Car Wink a voice command. Something like: “Car Wink. Thank you.” The little device displays an emoticon, like a thumbs up and the word “Thanks.”
Basic Concept: I like it! This is a novel and fun new use of voice. And, I’m all about getting more civil with your fellow drivers. Two thumbs up.
But, as always, the devil is in the details. How is this getting built? Let’s see…
Technical Details: After some digging around on the Kickstarter site, it looks like Car Wink is going to run off a mobile app which will be using Siri APIs on iPhone and Google Voice APIs on Android. Their FAQ’s say that they will be “using cloud computing technology.” So, of course, if it’s going to be doing recognition in the cloud, it’s going to be dependent on network connectivity. That might be problematic. The network’s great where I live, but it’s not great everywhere.
We also know this app is going to be used in a pretty noisy acoustic environment (a car). So right away there are two very significant factors (network, acoustic environment) in the success of this app which are not going to be under the control of the Car Wink team.
Now, I personally have had a lot of success with voice recognition using Siri in my new car — a 2017 Jeep Grand Cherokee. It’s got a bangin’ mic (people tell me my car calls sound great now, which makes life so much easier). Also, this car is sooo quiet. That’s turned out to be one of my favorite things about the car. And, it’s also got a button for one-touch launch of Siri on the steering wheel. As a result, I think Car Wink would work pretty well for me in my current car.
But my previous car did not work so well for voice. The mic and acoustic environment was terrible. “You sound like you’re talking to me from inside a tin can!” my mother would always complain to me. I was never able to get Siri to work in that car. And I didn’t have that neat little button integrated into the steering wheel. This sort of thing will play directly into the success (or failure) of the Car Wink user experience for the majority of drivers.
As far as network availability, they should probably use offline speech recognition (not cloud) in Android, and use something like Open Ears to do the same thing on iPhone. That would make the app more reliable in places where network coverage is poor, or drops out.
A bad acoustic environment is tougher to overcome. If they pick the keywords to choose the emojis carefully, that’ll help. I’d try to use keywords that start with hard consonants (to give the recognizer a chance to kick in). Also longer keywords (to give the recognizer a chance to understand what you’re saying).
Also it looks like they’re planning to have the visual display of the app itself as a backup, so you can choose your emojis directly from your phone. It’s sort of a good idea…. except you don’t really want to encourage people to be looking at their phone in the car.
Internationalization: Looks like they’re planning to ship everywhere in the world, right away. They have lots of backers from Singapore, Thailand, and Malaysia. Hmmm…. That right there says “Danger, Will Robinson,” to me. Internationalization with speech recognition is tough. The FAQ says English is the primary language, but they’re going to support others. It doesn’t say which ones or when, though. And if they’re planning on having non-native English speakers try to use the English speech recognition — well, they’re going to have a hard time with that.
Admittedly, Car Wink will probably have a small vocabulary, so that’ll improve the chance of getting a good recognition result. But we’ve seen this before — a certain friendly robot was initially supposed to be available internationally and that product team had to back off that goal. Glancing over the Kickstarter page, I’m just not getting the sense that the Car Wink team has thought about these voice related problems.
I’ve said before and I’ll say it again: “Sprinkle a little speech on top” doesn’t work. If you’re designing an application which is going to use voice recognition, you need to bake that in from the beginning. You need to have people on your product team who know voice. You need to recognize that voice is a new and different user interface. And you need to be prepared to take on those challenges, from a design, usability, and speech recognition point of view.
So, back to Car Wink. Would I back it? Maybe. I’m kind of a gadget fan. Plus, it’s a fun concept! Sharing the love on the road? what’s not to like? But I’m just not sure the voice experience is well-baked here.
Voice is not the lack of a user interface. Or, to get out the double-negatives, voice user interface does not mean no user interface. If you don’t understand that, it’s going to come back to bite you.
Or, to riff off a tagline from Big G (Google, that is): Voice first, y’all.