Should our digital assistants speak before being spoken to?
Digital assistants such as Amazon Alexa, Google Home and Siri are very useful, but there’s a catch: They generally don’t provide information unless you ask for it. That approach only solves half of the assistant equation. A truly effective assistant should speak up on its own, based on time, place or events to provide better contextual information.
Lights aren’t ideal notifications for voice assistants
Earlier this week, I shared some smart home automation ideas and one of them was to connect a scene with a calendar event. That go me thinking about how calendar events work on my phone: If configured appropriately, I see a reminder a few minutes before the event. And if there’s travel involved, Google Assistant pops up a suggestion on my phone to leave at a certain time based on current traffic.
This is super helpful and having a little pop-up or sound on the phone makes sense. But this simple approach doesn’t make as much sense to me on smart home assistants where you might get a visual notification. In the Amazon Echo, the light ring will light up in different colors and on the Google Home it will light up the little multi-colored ring on the top. However, while a buzzing or beeping phone in a pocket will surely get your attention, blinking lights on an Amazon Echo or Google Home might be missed.
While home assistants can sound a notification chime, I don’t think that’s the answer. Since the main communication method to and from these devices is voice, wouldn’t it make more sense to have the home assistant speak these event notifications out loud? I think so, although one could argue that they don’t want to hear from an Echo or Home unless you speak to them first. I get that. There should be a “do not disturb” option for these type of audio notifications.
Voice notifications are a natural fit for home assistants
Proactive voice notifications would add more contextual value to the experience and take advantage of the main interface embedded within these devices that have speakers. And it would make that experience more natural. If I have an event that I’ve forgotten about and my wife reminds me it’s time to leave, she doesn’t flash the lights on and off or make a chime sound. She provides me the actionable information I need at the right time.
Maybe I’m expecting too much from what a digital assistant can or should do.
I’d like the option to tie information from my Canary home security system, for example, into an assistant. Using geolocation data, the Canary keeps track of when everyone in my house leaves or arrives home.
I’m often upstairs in my office wondering if anyone is coming or going in my home but I can’t find that out without opening the Canary app and looking at the event timeline. I’d rather have the option of my Echo or Home simply say “Barb is home” when the system detects her arrival, for example.
Another example would be an automation I’ve manually set up and used in the past. If the lights are still on in the home office at 11pm, my smart lights remind me it’s getting late and that I might want to stop working, reading or watching Netflix. At the right time, my lights either briefly dim or I have a colored bulb turn red. I’ve had to create this automation with a workaround Python script built off my ambient notification code because the assistant sitting nearby doesn’t have the capability to gently suggest I call it a night.
Text-to-speech v. natural language processing
Frankly, it’s almost easier to implement these types of proactive voice notifications than it is to handle audio requests. The former uses TTS or text-to-speech technology for an assistant to speak while the latter uses NLP or natural language processing for an assistant to hear our requests. What’s for needed for TTS notifications is the data and a way to get that data from an app, sensor or device to the assistant.
I know that both Amazon and Google are providing some basic notification tools to developers and app makers, but they seem constrained to me. They simply don’t go far enough, at least for the future I envision and want. And while I may be in the minority here, I’m not the only one looking for assistants to actually assist more in this way: A Samsung SmartThings community forum has a number of folks devising their own methods to make such proactive voice events happen.
While not everyone wants their Echo or Home speaking before spoken to, I do think it’s the next step towards a true smart home assistant. Tell me what you think: Is this something you’d want in your house?