First off, I remain somewhat leery of the overall impact of voice command on the reality of day to day living. I know for example that there are specific voice geared devices available now and they are significantly better than they were four and five years ago,
I actually played with, and have played with each of the native voice control appliances, dragon naturally speaking and then with the three primary voice systems on competing hardware. I will say that the overall integration of Cortana with a Windows PC remains far superior to either the Amazon Echo or the Google Home.
Let’s talk visuals first. By visuals I am talking about the look and feel of the native devices. First off Google Home, it is without a doubt the coolest looking of the three. The Amazon Echo is the least visually appealing but at this point is the most media friendly of the three. The Hololens remains the coolest looking high tech device on the market.
Google home responds to the command OK Google. Amazon Echo responds to Alexa, and then the command. Cortana is the command (Hey Cortana) to engage with Cortana on whatever device you are using but in particular to drive the Hololens.
All three are limited today. Cortana remains the best cross platform and most functional on the Windows PC by a large amount. The worst is the only one that doesn’t have dedicated hardware today, Siri from Apple. You could argue that the iPhone is dedicated hardware for Siri, but if that is the case Apple has a long way to go, Cortana is better on the iPhone than Siri and frankly so is Google at this point.
Noise obviously impacts the ability of the system to interoperate. Cortana has the most power on Hololens and Windows PCs, frankly it has the most power on the iPhone as well today. But it is still limited by ambient noise.
As a technologist, I believe in looking ahead. I do not believe in selecting one solution and going down that path only. So, I have played with all of the devices on purpose. In part because in order to see where each system is weak, you have to see the competitive solutions.
Using the decibel meter on my iPhone (let’s give it a +/- 5% real noise rating) I set-up the following experiment.
My radio playing a shortwave signal (99% static) and the radio was set to a volume of 14 (the radio has from 0-20 as settings. 14 felt like it was loud enough to be heard but not loud enough to cause the rest of my family to come downstairs and complain about the noise).
Each of the primary systems was placed within 5 feet of the radio producing noise. The room, other than the noise of the radio is well insulated (home theater) and there is little to no ambient noise to cause further issues. I also did this experiment with the Sonos speakers in the home theater specifically playing talk radio.
1. Google Home commercials cause google home to respond to the TV. The product has a not complete feature set so I tested Google Home playing music of my choice. With the background radio noise or the background talk radio noise the Google Home was able to correctly hear commands 8 times out of 10 and to be sure I did another 10 commands overall Google home’s accuracy was 15 out of 20 commands correct. If you took he volume of the radio to 16 the success rate dropped to 8 out of 20 correct.
2. Alexa, the Amazon Fire and Amazon Echo control voice fared better with the talk radio noise than Google completing 8 of the first ten and 16 of the 20. Increased noise as with the Google home dropped its success rate to 9 of 20.
3. Cortana with Hololens ran at 18 of 20 in a noise filled room. With a windows PC, it was closer to 17 of 20. Still better in both cases than either of the others. Its rate was decent with higher noise 11 out of 20.
4. Siri struggled and ultimately was only able to correctly operate 14 out of 20. In louder noises, however Siri still maintained a better rate of response than either Alexa or Google home at 10 of 20.
5. Dragon did the best of any of the systems but it isn’t a fair comparison because Dragon has a microphone and earpiece located on my head. So the ambient noise has less impact overall.
If there was a talk radio commercial that activated any of the four devices that ended the command that I issued and started the new command. Overall Cortana remains the strongest voice recognition device but it is a risk when you add noise.
I am waiting for an artist to produce a new sample song. Hey Alexa have you ever said OK Google? Hey Cortana Alexa wants to meet you. OK google meet Alexa. And so on, then I can just play that song as I sit in my office and watch them all freak out.