Voice recognition works particularly well when the computer knows what you are trying to say, and just needs to track along with you.
Which leads to an idea for a plug-in for commenters – not for this web-site, but for some others that I frequent.
Before leaving a comment, the system will require you to turn on a microphone, and read some text out aloud, while the voice recognition system tracks along with you to the end.
What text will you need to orate? The complete article you are about to comment upon.
That way it will be impossible to add comments without having actually read the passage you are commenting on.
The Internet need not thank me personally for this idea; the improved quality of comments will be thanks enough.
Comment by Aristotle Pagaltzis on December 16, 2011
From the title, I thought you were going to talk about some anti-spam scheme.
And then I found I was right.
Comment by Julian on December 16, 2011
Excellent point, Aristotle. I guess that means checking the recording at the server side, and an inevitable arms-race between speech-synthesisers and speech-synthesizer-recognisers, although maybe the former would help the state-of-the-art?
Comment by Alastair on December 17, 2011
tl;dr
Comment by Andrew on December 18, 2011
Google would do it just to get a few billion voice samples of different people reading the same text. A couple of years later they’d bring out an e-book reader where instead of choosing some named voice, you’d just pick a spot on Google Earth and it would read the book in the local accent.
(or if you have gmail, it reads it out as you).