Slashdot Mirror
Gloves Translate Sign Language Into Auditory Speech
Zothecula writes about some pretty cool sensor gloves. From the article: "Since beginning in 2003, the Microsoft Imagine Cup has tasked students the world over with developing technology aimed at solving real-world problems. In this, its 10th year, students were asked to build their project around a specific Millennium Development Goal ... The winners have just been announced ... [and winning] first place (and US$25,000) in the Software Design category was the Ukraine's quadSquad with their EnableTalk gloves that translate sign language into speech in real time."
Another $20 an hour job gone...
Until I see a gorilla using them to talk about a secret diamond mine, I'm unimpressed.
"In God We Trust, All Others Pay Cash"
This is a cool idea. But it kind of only goes half way. Now they need a Thing that can translate spoken speech into glove movements. Oh, yea, baby!
"Although the software was developed under Windows Phone 7, the team was forced to turn to the older Windows Mobile platform for their entry because Windows Phone 7 doesn’t provide developers access to the Bluetooth stack, which is how the gloves communicate wirelessly with a mobile device running the translation software."
And that tells me all I need to know about WP7!
After watching the video, it seems that what they've done is create gloves which recognize the various fingerspelling signs. If somebody wants to sign "I need to withdraw money" (like, at a bank), what this allows them to do is to make the sign for "I", then "N, E, E, D", then "T,O", and so forth. Then the gloves feed that output into a TTS system. This works (because ASL users and English speakers share a writing system), but is horribly inefficient, and would be equivalent to a translation module that makes you speak every letter of the written words individually before putting the words into Spanish.
This is fundamentally different from "translating sign language", where the gloves would recognize the (much more complex and spatially oriented) sign for "I", for "need", for "withdraw" and for "money", and then translate that into "I need to withdraw money" and speak it aloud. Adding in the fact that ASL syntax is fundamentally different than in English, it's quite a tall order. Interpreters need not fear.
This is cool, nobody's denying that, and for some jobs, this might be great, but at the moment, I don't see it working much faster than taking out the requisite smartphone and writing down what you're trying to get across.
This actually does solve a real world problem and I commend them for accomplishing as much as they did. Often people who communicate via sign language are able to understand what other people are saying but they are at a complete loss in having others understand them. This bridges that specific gap in a way that is convenient for both sides.
Nothing to see.
With a text to speech app?
Slashdot's rate-of-post filter: Preventing you from posting too many great ideas at once.
The gloves translate dactylology (finger spelling). That's fairly easy, but it's horribly inefficient.
'Real' sign languages (like ASL) are much harder to translate because they are somewhat non-linear. A single gesture can describe several things at once: size, direction, emotional state, etc. There's no way you can translate it without fully understanding the context of the speech. And we all know how good computers are at such tasks...
Wow! That's new! Speech never used to be "auditory" before. That's a real breakthrough. Maybe these guys can tackle the old problems of Visual Sight, or Sensory Touch, or even ... Olfactory Smelling.
"Cock Up Your Beaver" does not mean what you think. This sig is intended to clog filters and annoy do-gooders
It's been 17 years already!
A grad student at Stanford at the Center for Design Research did this in the mid-1980s. It had to be connected to a workstation back then, of course.
Mod parent up! Finger Spelling is *not* Sign Language. If all this does is translate finger spelling into synthesized speech, the same thing could be done much faster and cheaper by just typing the words on a standard smartphone device.
This is not even cool. It is just, plain, wrong in so many ways. All of the money and hype spent making and marketing this device would reap 10X as much benefit if the same money were spent educating people about the real nature of deafness and sign language. The developers of this waste of time could start by taking a class about deafness themselves.
The fact that Slashdot perpetuates the inaccurate headline equating finger spelling with sign language just demonstrates how ignorant we all are.
the gloves say "Please fondle my buttocks".
Close palm, raise middle finger. Universal language at it's best.
Then I guess it translates sign language. Finger-spelling is just an alphabet. I don't know about other sign languages, but ASL is a full language with its own grammar and conventions, and it would take a lot more than a glove to interpret it. Positions of hands with respect to the body are important, as are facial expressions, and ASL's pronoun system is largely spatial with the handshape only indicating the type of pronoun (e.g. personal vs. possessive). Even if a piece of technology could reliably capture and interpret all that information, you're still back at the problem of natural language translation into the destination language.
Still, a cool hack.
They reward something basically worthless. All this glove can do is translate letter-by-letter "finger spelling" which is NOT ASL... it's finger spelling. I know virtually NO sign language, and I can basically read finger spelling. So... Microsoft gives someone 25,000 dollars for making a device that does what any decent American education can, in about a day. Yeah, I'm saying the common, empty-headed Twitter-reading, Facebook-using American idiot child can memorize 26 symbols made with fingers in a day. Most already know several, demonstrating their abilities to learn them... signs such as "the Bird", (meaning more than just one letter, it indicates general displeasure with someone or a group, and the hopes that they will soon be forcibly penetrated, or have an otherwise similarly bad day,) and "the Shocker," (expressing a general approval of the idea of manually stimulating the vagina and anus simultaneously...), "the Love" sign, (thumb, index and little fingers extended, middle and ring fingers curled down toward the palm, indicating affection for someone or something, or just approval of the idea of love itself,) "the peace" sign, similar to "the Shocker" except little finger curled down, and index and middle fingers spread... this is also commonly the victory sign, and is similar to a sign that indicates the same meaning as "the bird" in other cultures, mostly in Europe, although the other side of the hand is show.
If you want to impress me, make a pair of gloves, or some sensors worn on a hand, that can translate REAL ASL, not just 26 letters... or better yet, stop addressing, as is so common in the world today, the minor problems, and get on to the major ones, like figuring out how to enable people who might have to use ASL to hear and speak like everyone else on the planet. I am sick and tired of hearing about people throwing up their hands in defeat, in a sense, and trying to make the lives of disabled people better, by building a better wheel-chair, for example. Screw the better wheel-chair, focus on figuring out a way to let them WALK again!!! You want to help people, stop making ways to make minor, incremental improvements in their lives, and ACTUALLY HELP THEM. We have a society in which stem-cell research is controversial, but yet every new building that goes up has to be wheel-chair accessible. I heard on NPR that public swimming pools around the country are at risk of closing because some disabled people can't get in them because they don't have some ridiculous contraption allowing them to get into the pool if they can't walk or whatever. Instead of forcing people to spend millions or billions of dollars to accommodate a few people who mostly won't use those accommodations anyway, just so they can pretend they're not disabled, while at the same time making something people will probably misuse, or abuse, (kids will play with these things...) why not spend that money on helping them walk again?
This society's priorities are completely ass-backwards. There was an article in The Onion joking about the President signing the "Americans with No Abilities Act"... I sometimes wonder how far off from that moment we really are.
Having watched their video presentation on YouTube, this is downright offensive.
You know they don't work closely with hearing-impaired people, because the video is not Close-Captioned (oops: this presentation is not for the intended users); they also misrepresent the scope of the prototype: to recognize signs for individual letters is one thing, to translate from a sign language with its own grammar rules and idiosyncrasies is another.
The punch-line is that they say they use a Microsoft text-to-speech framework, as if that was somehow an enabling factor for their technology. Of course, it's mighty good that Windows text-to-speech is usable for their purposes, and they didn't have to turn to one of the half-a-dozen third-party frameworks.
Big Deal this was done already. http://www.wired.com/gadgets/miscellaneous/news/2002/01/49716 the kid did texting but that could just of easily been speaking. I call prior art on this one! And this kid also got a 103,000$ scholarship out of the idea! (from 2002)
Now lets just make the price affordable. Hey were is the Kickstarter on this!
Sucks for someone still learnign sign language.
"Tom, regarding your point about source control
(you accidentally scratch your ass midsentence)
you're an ass --
(you try to cover up this interpretation by coughing into your fist)
you cocksucker! cocksucker! cocksucker!
(you turn beet red, cover your eyes in embarrassment)
I can't even look at you.
(try to salvage the situation)
Tom, I am so, so sorry! Sorry! This was a communications problem. I mean to say, your idea is very, very good, so
(give thumbs up sign)
up yours!
What kind of WPM can be achieved with dactylology? Could this evolve into a replacement for keyboards?