Voice Actions, Siri’s mentally challenged little brother
Siri, Apple’s new iPhone 4S-only voice control system, has gotten a lot of praise since it was announced. Many people found that quite annoying as voice control isn’t exactly new, and has existed as both native apps and third party apps on various mobile OSes for a while now. After doing some research I would an app that is available for both Android and iOS that promises to do a lot of the same things. Seeing as how my identically-specced-as-4s-iPad was not given Siri with iOS 5, I decided to give Voice Actions a try instead. Hilarity and headaches ensued.
Voice control is hard, because language is complex. Even if you ignore dialects and the fact that English doesn’t sound the same from one country to another, we humans don’t exactly make much sense if you really analyze what we’re saying. One of the examples that Apple showed off was Siri’s ability to understand that we want to know the weather when we ask if we need an umbrella. You might argue that “I don’t know what you need” would have been a more appropriate answer from a literal point of view. The point is though that voice control is so much more than voice recognition because the software needs to understand what we mean rather than what words we’re saying. That’s why Siri requires a connection to Apple, so that their servers can be constantly updated to help it understand us.
Voice Actions is a perfect example of an app that has the voice recognition bit (sort of) in place, but not the voice control part. I played with the app for a while and the number 1 annoyance was not that it didn’t get all the words right, but that it didn’t understand what I was trying to do. When compiling an email, I though it natural to tell it to “delete that word”, “delete the last sentence”, “scratch that”, “go back to the last sentence” and similar commands. What that resulted in was emails full of “delete that word” in the actual text”.
Here’s an example:
Me: “I’m trying out an alternative voice recognition system for iOS”
Voice Action transcribes: “i am trying out the normal time at the voice recognition system for ios”
Me: “alternative voice recognition system not normal time.
Voice Action transcribes: voice recognition system normal time
Me: “NOT normal time”
Voice Action transcribes: my normal time
Me: “NOT normal time”
Voice Action ranscribes: lots normal time
Me: “N. O. T!!!”
Voice Action transcribes: not
Me: sarcastically to the app “thank you”
Voice Action transcribes: thank you
Full text entered into email: “i am trying out the normal time at the voice recognition system for ios voice recognition system normal time my normal time lots normal time not thank you”
I also had a lot of fun trying to make it understand what feature I wanted to start. “Remind me about my dentist appointment tomorrow at 10” actually made it ask if I wanted to be reminded tomorrow at 10 AM, but when I said “yes” it didn’t put anything in the calendar. I repeatedly tried to make it “email Grahm” which only started working once I stopped saying “Grahm” and started saying “Graham”. One time it even started doing an email to Grahm when I had told it to email Graham, but it had heard it as “email Greg” (according to the on-screen text). Then there were the two times it asked if I wanted to call someone after I had told it to do an email.
So, as you can see, Siri might have something going for it after all. These types of features might already exist, but seem to be a lot less advanced. I also took a look at some videos of other, non-iOS options out there and they all seemed to fall short of the Siri demonstration – though I can’t be 100% sure on that without having tried both Siri and those.
What’s very clear though is that Voice Actions is not the way to go. It’s also a paid app, and $6.99 is at least $6 too much for something that doesn’t work better.
[iTunes]
