Slashdot Mirror
DARPA Starts Ultimate Language Translation Project
An anonymous reader writes "Defense Advanced Research Projects Agency (DARPA) has launched the ultimate speech translation engine project that would be capable of real-time interpretation of television and radio programs as well as printed or online textual information in order to be summarized, abstracted, and presented to human analysts emphasizing points of particular interest." If combined with the tower of babel project we discussed earlier, it could only lead to awesomeness.
If you consider that now the government will be able to spy on you in your native language to be awesome, then I suppose giving the Feds this sort of technology can only lead to awesomeness.
Surveillance of civilian populations under the guise of "monitoring terrorists" is not something that I'd consider awesome. Irksome, yes. But not awesome.
Just feed this new system a few reruns of Japanese television game shows. After that, we will be safe from automated snooping for at least another decade. As a plus, all artificial intelligence projects at the DARPA will be set back by another decade as well.
[
Seriously though, I just don't believe it. I've worked on a number of DARPA robot projects, and have heard a lot of their babble. They claim to be funding all these fantastic ideas, but none of them ever work except in a limited capacity. The robot projects I worked on were very lame in that DARPA created these really specific environments for the robots that were light years away from what they were saying they were really going to do. All of the Universities involved failed to accomplish even the simplest tasks. So my experiance with them is that they talk a big talk, and no one ever goes back to check "hey did you really ever do that ?" Now granted, some of their work is supposed to be high risk, but they never emphasize which projects are expected to have a high failure rate. Largely because they don't care. It's really all about funding your academic buddies or whoever is going to be able to scratch you back in some way. It is very much an old boys network, with an emphasise on PR and not much about real science. Much like the MIT media lab. (Just thought I'd get another jab in there....)
One of the problems with using humans is that they are expensive--the other is that they become bored easily. It isn't like the defense establishment isn't using human translators, the NSA is the largest employer of translators in the world. They use humans in every listening post out there, but for the same reasons that humans make lousy airport security sceeners, they make poor translators AND intelligence analylists. This isn't saying that machine translators are a panacea, but they can solve a small section of the problem that we have been trying to solve with a very human capital intensive solution for years now.
This project, along with CMU's Tower of Babel, certainly get props in the coolness category, but the practicality is still lacking. I believe DARPA is barking up the wrong tree for now, or at least biting off more than they can chew.
Speech Recognition is the hardest problem to tackle on the path to recognition, and MUST be addressed before there is a viable real-time (or even delayed) translation engine. Currently, even the best speech recognition software can achieve at best ~80% accuracy when faced with a large vocabulary with no limits on speakers/dialects, and this level of accuracy is typically not achieved in real-time. While this 80% level is actually pretty good when transcribing to text (since the reader can typically decipher what the computer meant), it's downright awful if trying to translate the resulting text to another language.
For example, if I say "I like ice cream" into voice recognition software and 'hears' "I like, I scream", the reader might understand what this means, particularly if they say it in context and aloud. However, let's say we translate each sentence into Spanish ("Tengo gusto del helado" and "Tengo gusto, yo grito" respectively, according to Babel Fish), and the speaker would be completely lost as the out of context phrases don't sound anything alike. In a natural language translation, even under relatively accurate recognition scenarios, would be frought with misunderstandings.
Once speech recognition is tackled, it's just a matter of translation then voice synthesis. Fortunately these problems aren't nearly as difficult, and current solutions would suffice (with the only pitfall being poor grammer in the destination language, and a robotic sounding voice).
Crack - Free with every butt and set of boobs
To properly translate all the nuances of some languages actually requires a lot of skill, and sometimes translating can be ask much interpreting as anything. Granted, this is something a human could handle better than a machine, but the problem is that humans also have a bias. Yes, there have been cases wherein human translation has caused problems because of bias or even due to being outright wrong.
I reminds me of the old joke:
Guard: Now tell me where you hid the money, or you will suffer
Translator: Tell him where the money is, or you will suffer
Prisoner: I'll never speak
Translator: He says he won't tell you
Guard: *putting gun to prisoner's head* Tell him I will blow his brains out if he doesn't tell me immediately
Translator: He will shoot you in the head unless you tell him now
Prisoner: I buried a million dollars under the floorboards in the old woodshed
Translator: *pauses* He says you don't have the guts to shoot him...
Seriously though, I just don't believe it. I've worked on a number of DARPA robot projects, and have heard a lot of their babble. They claim to be funding all these fantastic ideas, but none of them ever work except in a limited capacity.
..." and translated it as "I absolutely would mind to tell you about ..." which is the exact opposite. Many languages, such as Russian, Spanish and Portuguese (and no doubt others) use double negatives to express negation. "I don't know nobody" is quite correct in Russian, Spanish and Portuguese although it is quite grammatically incorrect in English if your intention was to say "I don't know anybody". Programs that translate into English from languages that use double negatives often fail to correctly translate the negation. Maybe there are some that get it right, but I've never seen any. Text translation programs are very poor at distinguishing between words that have uses as different parts of speech. Here's an example:
This is a big pipe dream that is extremely unlikely to work any time soon. How do I know that? Right now, I think it would be reasonable to conclude that computer technology today is good enough to do accurate text translation. Can it? Well, it depends on how picky you are. There are always mistakes, sometimes glaring ones, in text to text translation programs. I can speak Russian and for convenience (to get quick rough translations) at one time I owned what is probably the best Russian-English text translation program. It's much more accurate than Babelfish. It still left a lot to be desired. It would be about 80-90% accurate, but no more. I remember one time when it took a statement in Russian that said "I absolutely would not mind to tell you about
She sings like an angel.
In this sentence, "like" is an adverb, but it can also be a verb ("She likes to go shopping."). A text translation program might fail to correctly understand that "like" is an adverb here and say something like:
She sings and angel is pleasing to her.
I could give a lot more examples, but these are enough. If we can't even do a better job right now at text translation, how on earth is DARPA going to get speech translation right? This is the kind of project that gets funded by idiots who have never studied foreign languages and believe that the Star Trek idea of a Universal Translator is only a few years away.