Today, Google is shipping a major update for Google Home: multi-user support. Up to six people will be able to connect their account to Google’s voice appliance, which will then be able to identify each person just by listening to them speak. The Google Home app jumped the gun on this launch earlier in the month, but now the feature is actually rolling out.
After connecting their accounts, each user will have to go through the usual hotword training, reciting “OK Google” to the device three times. For future commands, Google Home will then call out the name of the person it detects before it gives a response. So if you ask “What’s my day like?” Google Home will respond with “Good Morning [name],” and then rattle off the weather and calendar appointments. To start the setup process, look for the card that says “multi-user is available” in the Google Home app. Thanks to the way Google rolls out features, it might take a few days.
On-the-fly detection of voices sounds like a challenging problem to overcome. During the training process, your “OK Google” phrases are analyzed by a neural network, which Google says can “detect certain characteristics of a person’s voice.” Future “OK Google” commands are compared to the previous analysis, and Google Home decides if it’s a match. Google notes, “this comparison takes place only on your device, in a matter of milliseconds.”
We asked Google how confident it was in its ability to distinguish users only by voice: “We’ve worked hard to make the Assistant work well for you, and we’re excited for users to start hearing their music, news and more. With that said, we’re just getting started and we won’t be perfect. We don’t recommend that users rely upon voice identification as a security feature.”
Unfortunately, this doesn’t solve the biggest problem we’ve had with Google Home since launch: all the missing commands compared to the Google Assistant on a smartphone. A $130 dedicated voice appliance should provide Google’s best voice experience possible, but Google Home still doesn’t support reminders, creating calendar events, or taking notes. It also can’t do a lot of phone stuff like sending messages or making calls. Android devices work with Google Home to give it priority for all voice commands, so even if you have an Android phone capable of handling these commands (anything running Android 6.0 and up), Google Home will intercept the voice command and say “I can’t do that yet.” If you just want the most voice capabilities, you’re better off unplugging the Google Home and letting your more-capable phone handle everything.
Many of the missing features involve writing data to personal accounts. When we heard multi-user support was coming, we speculated that would be a key enabler for the missing commands. Google views Google Home as a public device, so it doesn’t want just anyone to add reminders and calendar events to your account. Now that it can identify individual users, it should be more willing to enable these “personal data” commands. We asked if those were coming soon, and Google told us there was “nothing new to announce today beyond the ability for multiple users to connect to one Google Home.”
The user detection should at least stop random recordings of “OK Google” from triggering Google Home, meaning Burger King-style hijacks won’t be possible in the future. Google says this new feature is rolling out in the US today and will come to the UK in “the coming months.”