On Apple’s Voice Recognition “Assistant”

September 30th, 2011

When I mused on the iPhone 5 / 4S a few posts ago, I mentioned NFC as one wildcard — well, another I didn’t think of at the time is Apple’s long-rumored voice recognition assistant.

Voice recognition has long been a computing holy grail; I’m sure any tech-head has dreamed of saying “Earl Grey, hot” and having a computer respond appropriately. But I’ve long been disappointed by voice recognition in OS X. Every year or so I fiddle with it and it doesn’t work as well I hope. And so I have a short list of what I’d need from voice recognition before I’d use it regularly:

• Must understand normal speech in order to get results. If I have to change my speaking, as in, adopt a monotone, robotic, perfectly enunciated voice, it’s not worth it.
• Similarly, must be able to understand human grammar. Having to remember a correct but unnatural sequence of specific words is too complicated.
• Faster than typing. If one needs to go back and fix spelling mistakes it’s not worth our time; may as well just use the keyboard.

And so this fictitious “Assistant” video from AppleInsider is great idealized representation. A phrase like “Send a text to ‘x’” sounds totally doable. I’ll use “Assistant” if it works like this.

Will “Assistant” work this well? This is where I have faith in Apple’s focus on ease of use and goal of perfection. They’ve likely been working on this for so long because it’s got to be right when finally released. They know mainstream users like myself will try the feature once or twice, and if it doesn’t work as expected — never try it again.

Therefore, if it’s announced on Tuesday I predict it will work really well. Nobody wants to return to the “egg freckles” days. We want the Knowledge Navigator.