Tuesday, December 8, 2009

Speech Recogition

There are programs out there that allow you to control your computer by dictating into a microphone. That's awesome. You can also use this to dictate lengths of text, saving keyboardists gajillions of keystrokes. As computers get smaller, it's possible that portable computer will need to use this technology, because the computer itself will be smaller than is practical to install a keyboard to. I already own a device one quarter the size of my keyboard, intended for embedded use in, say, a factory to measure probes. This is the good news.
The bad news is that the technology is both expensive, and never exactly sure of what you say. You. Have. To. Talk. Like. This. For. It. To. Understand. You. It eliminates the less likely expressions. "Has he got, uh, a flight to Hamburg" is more likely than "Izzy god hat fight meat barge." It has to remove stammers (such as "uh" and "hmm" and the like.) and deal with some irregularities of pronunciation.
For instance on that, ask an English speaker to say "prime minister." They will usually say "pri minister." The common "m" sound smears together. One of the best speech comprehension tools was invented by a Chinese man who learned English very late in his life, because he had to learn these intricate rules himself, and had more an intellectual than instinctual understanding. It would be nice if I could remember this intrepid researcher's name, or where I read this from.
In any case, some day there will be a computer the size of your fingernail that you operate by talking to, and it will report back by speaking with a mechanical voice (speech synthesis is very sophisticated at this point), causing a massive increase in people who seem to be talking to themselves in public.

No comments:

Related Posts Plugin for WordPress, Blogger...