In modern day society, every yearns for things to be easier, to be more efficient, and convenient. When faced with a task, the first thing that comes to mind is "what's the best way to approach this?" Modern inventions make what used to be time consuming tasks seem less like a burden and more like a chore.

Everything has evolved into what seems to make the most sense, and computers have been no different. First we started out writing by hand, with a pencil. Then typewriters came along and we started typing with them. Someone came along and decided that typewriters weren't good enough for people, so he invented the word processor on the computer so that things can be easily edited. Still another person decided that he didn't like carpal tunnel syndrome so he came up with the idea of voice recognition software.

Well, I have a problem with that kind of software. I don't know how often this happens to other people, but the software just doesn't like to recognize my voice. No matter how long I "train" or "condition it for, typing up stuff is still more efficient than going back and editing all the mistakes that voice recognition produces. I propose a new type of input for inputing text: face recognition.

Most people have different tones of voice. That's where voice recognition fails. But almost everyone utilizes the same muscles in order to speak and make sounds. The lip reading software would require a camera or some sort of a three-dimensional scanner that recognizes what kind of muscles you are using in what order, and based on that, it will type out the desired word. It'll recognize more words with better results. That, and you could use this product silently so that you won't sound like some idiot talking to themselves.
-- cardeguy, Feb 29 2004

AVSR: lip reading software from Intel http://news.com.com/2100-1008-998576.html
"The Audio Visual Speech Recognition (AVSR) software tracks a speaker's face and mouth movements. By matching these movements with speech, the application can provide a computer with enough data to respond to voice recognition commands, even when these are given in noisy environments" [bristolz, Oct 04 2004]

A bit more on lip-reading software from Intel.com http://www.intel.co...mputing/sw04034.htm
It's open source. [bristolz, Oct 04 2004]

DoCoMo: "Lip-reading cellphone silences loudmouths " http://www.newscien...s.jsp?id=ns99992122
"DoCoMo's early prototype works out which words are being said by using a contact sensor near the phone's mouthpiece to detect tiny electrical signals sent by muscles around the user's mouth. The signals are then converted into spoken words by a speech synthesiser, or into text for a text message or email." [bristolz, Oct 04 2004]

It'll still have to deal with the continuity of flowing speech (as opposed to only; handling; separate; words) which I understand is a decent fraction of the trickiness of speech recognition.

But a camera combined with conventional speech recognition might be a good approach; I'd be surprised if someone hasn't tried this already. Certainly I find that, as a human, it's useful to watch someone if I'm having difficulty understanding them.
-- benjamin, Feb 29 2004


Baked in "2001: A Space Odyssey" sort of.
-- waugsqueke, Feb 29 2004



random, halfbakery