photo: Mrschimpf

Voice is the new interface, lacking some improvements

I received an unexpectedly reminder the other day from our Google Home smart speaker to resume this post. There I was, quietly watching a YouTube video about Nest when Rev Run happened to start a conversation with our GHM, and that got me thinking again about flaws in the usability of this relatively new interface to our smart-home.

Voice and speech recognition are rapidly maturing along with natural language processing, creating a shift from training users to interact with devices, to training devices to better interact with users. What better, more natural way to interact with another brain (be it biological or digital), than by using dialog? (Until we get to BMIs anyway, but that’s another story.)

While Siri has been a regular tool for the hands-free creation of reminders for many years, it took an always ready voice assistant in our kitchen to give our family a real taste of the technology. Our pilot run lasted 4 months with the Google Assistant SDK running on a RPi inside an old speaker box atop our refrigerator.

I obviously enjoyed building this new gadget and was excited to see how my family would embrace it, but had no idea what it would develop into. Turned out to be quite useful as a kitchen helper in the form of a timer, a shopping list manager, a unit converter. The kids also got a kick out of it, regularly asking Google to sing a song (birthday, Christmas and a selection of silly songs are “performed” by Google’s Voice assistant, though music services aren’t available in the SDK). We also controlled a bunch of our connected devices through it, and occasionally asked it to translate phrases.

I considered the pilot a success after 3 months, and thus started looking into buying a commercially available assistant, because the SDK had some important features missing. Two features on top of my wish-list were: music service and multi-user support. The first one is kind of obvious, but the later only became a need during everyday use. My Google account was connected to this Google Assistant SDK, and thus anyone issuing a command was essentially interacting with my account. Any requests regarding calendar, shopping list were directed at my personal account without any authentication, and were open to anyone who set foot into our home. I am fortunately not aware of any abuse of this openness, but it did leave me wanting some form of security through voice, and at the same time multi-user support, so each member of the family can access their own personal data, calendars, playlists, etc.

Google happened to release Voice Match around this time, and with the very affordable Google Home Mini, I was instantly sold on upgrading to the “real thing”. Little did I know that I’ll be slapped once again by the not-all-regions-are-created-equal fact of our digital lives. First slap came as I visited the Swedish Google store, and did not find the product, so had to order via a friend in Germany. S̶e̶c̶o̶n̶d̶ ̶s̶l̶a̶p̶ ̶c̶a̶m̶e̶ ̶a̶s̶ ̶w̶e̶ ̶t̶r̶i̶e̶d̶ ̶t̶o̶ ̶a̶d̶d̶ ̶a̶n̶o̶t̶h̶e̶r̶ ̶u̶s̶e̶r̶ ̶t̶o̶ ̶V̶o̶i̶c̶e̶ ̶M̶a̶t̶c̶h̶ ̶o̶u̶t̶s̶i̶d̶e̶ ̶t̶h̶e̶ ̶U̶S̶. Having written that sentence, I decided to investigate a little and chatted a few minutes with Google Home support. There was a less than obvious reason why my Wifey couldn’t link her Google account from her iPhone, namely that her iOS language was set to English UK, rather than US. Change. Reboot. Launch Google Home, and suddenly the option was available. Weeee!

Seems now we have access to just about all the cool features this voice assistant supports! That leaves me with two important usability items I would really like to bring to the Google Home dev team regarding Home Control (voice command for connected devices).

Pet peeve: Having to say the name of the room the GHM is in.

Current behaviour: Entering the kitchen, I have to say “Hey Google, turn on the kitchen lights”.

Preferred behaviour: Entering the kitchen, I just say “Hey Google, turn on the lights”. Or simply “Hey Google, lights”.

There is no real reason to add the name of the room, as each Google Home, along with every other connected device is assigned a room, and thus it is safe to assume that if I don’t specify a room, I am referring to the one we (human and GH) are in. Now you might be thinking, but then “what do I say if you want to turn off all the lights in my home?!?”. Simple, you make it clear that you are referring to all by saying just that, “turn off all the lights”. Considering the number of times people likely refer to devices in the room they are currently occupying versus all devices in every room of the house, I would argue it makes far more sense to make such commands local first, and only expand the circle of targeted devices when specifically asked to do so. This would also be a win when it comes to visitors who may not know what you happened to christen your beloved rooms. After all you may have felt inspired and refer to your kitchen as “crumb-stash”, making it a mere impossibly for others to guess and thus interact with those lights (lacking switches).

My second request to the Google Home dev team is specifically regarding guests. It would be handy, if not outright essential to have access control for devices. Locks, thermostat, security system should only be available to validated voices and absolutely not everyone who can manage to be heard by a Google Home. Similarly to how owners determine who gets to borrow the key to our physical homes, we should have a say in who can command our Google Homes, to avoid any funny business

My quick chat with Marc from the Google Home Support Team was rather positive and they seemed to appreciate feedback, so I sincerely hope they will take my ideas under consideration, and lead to better user experience in this new era of voice interfaces.