Run any TV audio feed into a device running voice recognition, and simultenously run the output through a translation engine and display requested language.
So you'll get a bit of broken telephone -- it'l make that Ingrid Bergman movie immeasurably more entertaining.-- theircompetitor, Jan 02 2004 Simultaneous Translation http://biz.yahoo.co...6/clw532.html?.v=12 [theircompetitor, Nov 02 2005] YouTube laucnhes spanish auto captions http://techcrunch.c...aptions-in-spanish/ [theircompetitor, Jun 16 2012] would be interested to see if this could work. I suspect that the delay would be far too long making the film unwatchable.-- jonthegeologist, Jan 02 2004 The delay should not be that bad especially with a dedicated device (i.e. not one running hundreds of other tasks). But with Moore's law even if it's slow today it won't be in a year or two.
The accuracy of recognition would still be the biggest hurdle, followed by accuracy of translation. But this is definitely demoable.
Now if you're really getting ambitious, maybe we can throw in text-to-speech and have simultenous dubbing into required language :)-- theircompetitor, Jan 02 2004 tc: you'd get my vote if you ran the audio feed under the specified language, then translated it back to English. I mean, exactly what is Ingrid Bergman saying in Swahili? jtg: in all fairness, if you're translating movies, you can easily do the translations in advance.-- DrCurry, Jan 02 2004 DrCurry -- so where is the vote -- it does say multilanguage in the title?-- theircompetitor, Jan 02 2004 I was trying to decide if that was what you actually meant.-- DrCurry, Jan 02 2004 thx -- I would think any to any language, so long as the voice recognition engine exists.-- theircompetitor, Jan 02 2004 So the idea is for a universal translator? How is this not magic?-- phoenix, Jan 02 2004 phoenix -- not magic at all. Results would definitely be broken telephone, at least at this stage of the game. But I think we're getting there.
Here's a recent related real-life example -- took a paper copy of a German language RFP (request for proposal), OCR'd it, then computer translated it to English.
Not Shakespeare, but definitely could read the doc.-- theircompetitor, Jan 02 2004 Phoenix, every part of this exists already - we have speach to text typos occure, launguage to laungauge (poor translations do occure), and text to speach (rather dry though).
oh, sorry theircompetitor.-- my-nep, Jan 02 2004 Jutta, it's not the same idea because I propose to feed live audio in as well into a voice recognition system, then translate its output.
Text to speech is definitely not ready for Finnish version of Hamlet, but I bet it won't sound worse then a Kung Fu movie.-- theircompetitor, Jan 02 2004 Dammit, if I think it, they will do it. Only took 8 years. Ok, people, I'm taking submissions :)-- theircompetitor, Jun 16 2012 random, halfbakery