/hosts is defeated using MacSpeech Dictate. Well, that was not very successful first sentence, was it?
This blog post is being dictated using MacSpeech Dictate. That’s more like it.
I am using the jawbone Bluetooth headset, connected to my Mac. Actually, the word was “A.” not “the” in the previous sentence.
In general, I am not a great fan of voice dictation software, which often has both gross mistakes where nonsensical words are returned and fine mistakes, where short connecting words are lost or transposed. This is all okay,. Sigh. This is all a pity, because it certainly seems like composition by dictation would be an enormous productivity boost for a writer.
I’ve been using voice dictation software for, let’s see, something in excess of 15 years. I first reviewed voice dictation software when I was working for the magazines, and I left in 1996. I think I first reviewed Dragon for AI expert Navy from (? — I have no idea what I actually said) 1992.
It’s Almost always been just like it is now, where as soon as you think “okay that’s an acceptable level of performance,” you get results like the last paragraph. You really can’t turn away from the screen, lest you risk some catastrophic recognition problem.
The other thing that is really frustrating is that you don’t realize how much composition is not stream-based. When the voice recognition software makes a mistake, navigating back to it then jumping to the end of the document has always been, in my experience, a frustrating experience. Given that there’s probably some amount of correction going into 80 percent of the sentences you write, whether because of transcription error or simply because you want to rephrase, any problems with navigation are a big deal.
I think that is as far as I will go with this blog post. Your mileage may vary.
Manual corrections for legibility in red.