Instagram Twitter GitHub Linked In

I keep reading that voice recognition has reached a better than 95% success rate but I’m yet to see it. Personally I’m more surprised when it works than when it doesn’t. This is across Siri, Alexa and Google Voice, though for me Google has a clear advantage. Admittedly I have an unusual accent coming from NZ but English is my native tongue, in fact my only language.

What makes it worse are the false positives, “OK” then nothing. “Alexa, lights off”, “OK”, then nothing. In the connected home I’m not sure this is always the fault of the voice platform of course, there are intermediary steps that fail but that feedback should be coming back. Giving a positive result when the action has failed is not helpful.

At this point I’ve given up on Alexa and unplugged my Echo Dots. Not only were the results mediocre but it was incorrectly activating when I hadn’t asked it to. Answering the phone then having Alexa burble away randomly at some point that it doesn’t understand what I said became an all too frequent and distracting issue.

There are times when voice is a natural interface to use. I do find the utility fairly limited however. The times I want to use voice are infrequent:

  • Control the lights
  • Set a timer
  • Set an alarm
  • Set a reminder
  • Listen to the news (this is the only functionality I got out of having a smart speaker in the room)

If the experience was frictionless I’d get the appeal but the mangled interpretations have become a source of frustration and one that I can avoid by simply not using. After all there’s nothing provided by a voice interface that I don’t have another option for that I know will actually work all the time.