- Announces Google calendar items and incoming mail
- Provides voice control for compatible devices, such as the WeMo smart switch
- Often misunderstands voice commands, with amusing but frustrating results
- Teaching it even simple tasks can be more complicated than it’s worth
The Ubi has too much trouble understanding voice commands to make it worth the time it takes to teach it how to perform tasks in the first place. But it could be fun for the patient hobbyist.
Its manufacturer ambitiously calls the Ubi—a voice-operated computer—the “Voice of The Internet.” It’s a small box (measuring 4.75 by 4.75 inches and 1.5 inches thick) designed to plug directly into an electrical outlet. With its integrated microphone and speaker, it looks much like an intercom. Once connected to your home Wi-Fi network, you can use the Ubi to control smart devices with voice commands, access online information and entertainment, get voice alerts for calendar events and incoming mail, and perform other potentially useful tasks.
UCIC says Ubi works with smart-home devices such as the Nest Learning Thermostat and a variety of gadgets in the SmartThings portfolio. I tested it with a Belkin WeMo Insight smart switch, basically a three-prong outlet you plug into an existing outlet and turn on and off via Wi-Fi. The Ubi also works with If This Then That (IFTTT) recipes, and the list of recipes available on the IFTTT site, some created by the company itself, gives a good idea of the range of tasks the Ubi can accomplish.
Finally, the Ubi can also work with systems that can accept HTTP requests. This capability is like a “light version of an API,” according to a company representative, and is aimed mainly at tinkerers who want to experiment with integrating other systems with the Ubi.
Once you plug the Ubi into its USB power supply, you complete the setup with your browser (Chrome recommended) or the UbiCC Android smartphone app. To start with, the Ubi creates its own Wi-Fi network that you connect to. Then you tell it what your own Wi-Fi network is and enter its password. The Ubi connects to your network, you reconnect to it, and you’re ready to go. (This, by the way, is also the way you set up the WeMo switch.)
Talking to the Ubi
The process was straightforward but not smooth. I was never able to connect to the Ubi network via Chrome on my Mac. I finally tried it with the Dolphin browser on my iPad 2, and then I was able to complete the setup process.
You get the Ubi’s attention by announcing “okay, Ubi.” (Generally, “okay” by itself was enough, which meant the Ubi often woke up during my telephone conversations.) At that point, a band of blue light begins glowing around the perimeter of the device, and when the light turns green after a second or so, Ubi is ready for a command.
The Ubi portal Web page provides a list of “things to say to your Ubi.” But talking to it was a very hit-or-miss prospect. I tried asking, for example, “How high is Mount Everest?” and it gave me the correct answer. But when I asked, “How many calories are in cornbread?” (which I happened to be making at the time), it replied, “[unintelligible] moved to San Francisco in 1999 and left in 2002.” It’s worth mentioning that the Ubi speaks in a robotic female voice (there is a command to “change voice,” but no indication of what it would change to, so I didn’t try it). There are no soothing Scarlett Johanson-esque Her interactions to be had here.
Those conversations were typical of most of my exchanges with the Ubi. I found the default voice volume too high, especially considering its robotic nature, so I wanted to make it quieter. Reading directly from the “things to say to your Ubi” list and speaking very slowly and carefully, I said, “Decrease volume.” The Ubi replied, “AC or DC depends on the power source.” I said, “Minimum volume,” and it actually turned the volume up and replied, “Volume is up.” I finally tried “can you lower the volume?” and it did and replied, “Volume is down.” According to the company, the Ubi is supposed to get better at understanding you the more you use it. I can only hope so.
Incidentally, you can see what the device thought you said and what it replied in a log accessible from the portal page. That can be kind of amusing: apparently at one point it thought I said “boss,” and it replied, “My best friend is myself.” From the portal page, you can also get environmental information such as temperature, humidity, and light level. I’m pretty sure the temperature measurement was wrong, though—there’s no way I was sitting in a room heated to nearly 80 degrees Fahrenheit.
The real power of the Ubi lies not in what it can do on its own, but what it can do when interfaced with other devices or tools. The company provided me with instructions for how to make the Ubi work with a WeMo (which I needed, because the procedure isn’t obvious). Once the WeMo is added to your network, its icon appears on the Ubi portal page next to the Ubi itself. First you click on the Ubi icon to bring up its own page, and then you click the Add Lesson button.
Ubi tasks take the form of a trigger leading to an action and, if desired, a subsequent action. For the WeMo, I chose the trigger utterance (other choices include sound level, temperature, or clock) and typed in the command I would say: “Turn on WeMo.” I then chose turn on from the popup menu of WeMo actions. I could have stopped there, but I added another action to have the Ubi confirm by saying, “WeMo is on.”
After all that, I was able to turn on my Christmas lights by saying “Okay Ubi. Turn on WeMo.” I would need to teach the Ubi another lesson to turn the WeMo off. I could perform the same actions using the WeMo’s own app and my tablet or smartphone without needing to program anything, so the only advantage Ubi offers here is the hands-free element.
I also added a premade IFTTT recipe to have the Ubi announce when I received email. That was a mistake. I get a lot of email, and having a robotic voice intone “you’ve got mail” and attempt to read the subject line every five minutes makes it hard to work. I finally turned the Ubi off, but when I turned it back on, it announced every message that had arrived in the meantime. It seems that while the Ubi is in the middle of doing something (like announcing two dozen emails), you can’t get its attention with the “okay, Ubi” voice command. I finally turned the device off again and then disabled the IFTTT recipe.
The inability to interrupt the Ubi was also a problem when I asked it to play music. The device is supposed to be able to play any song from SoundCloud or a Grooveshark account. A few seconds after it starts playing, the device will ask if it should continue, and if you say “yes,” it’ll finish the song.
First I asked it to “play Wilco,” and it started playing a song by some other band—maybe My Morning Jacket? Close, but no cigar. So then I tried “play ‘Poor Places,’” a specific Wilco song. The Ubi started playing something exotic that sounded like late Jefferson Starship with Arabic vocals. It was certainly interesting, but I couldn’t find any way to see what it was, and in any case it wasn’t what I’d asked for. And as with the email announcements, I couldn’t get the Ubi’s attention in mid song. When it got to a bad guitar solo, I once again had to walk over and turn it off. I’m still curious as to what the song was.
Other things I tried: I subscribed the Ubi to my Google Calendar and got it to remind me when I was supposed to put in my eye drops. That was the easiest and most successful interaction I had. I also added myself as a contact, typing “jake” in the box that tells it how to pronounce my name. Saying “email Jake” produced replies like “please rephrase that” and “try your request again with a contact name.” I changed the pronunciation to “bob,” figuring that had to be easy to understand, but got the same responses. I finally changed the pronunciation to “me” and after a couple of replies like “do you file your email?” it finally sent me a message.
Is the Ubi a worthy investment?
There are situations in which the Ubi would be useful enough to be worth the trouble of setting it up. The lighting-control scenario I described earlier is one. You could install an Ubi by the stairs to your attic or basement, so you could tell it to turn on the lights with your arms full. There may be some other compelling IFTTT recipes that I didn’t try. It could be a boon to the visually or physically impaired, once someone else sets up the actions. But the voice-recognition would need to be much better than it is today for someone to rely on it for any mission-critical applications.
At this point, the Ubi is an intriguing hobby for someone wanting to explore the possibilities of a voice-controlled smart home and is willing to put up with the equivalent of speaking to a child whose first language isn’t English. It also offers a glimpse of what might be possible a couple of years down the road. UCIC might have recognized that, because the company has reduced the Ubi’s price tag from its initial $299 to just $199.