Sign language can greatly improve the communication capability of hearing-impaired people, but there's still a major barrier in that most people don't understand it. New prototype gadgets could change that, by automatically translating hand motions into audible speech that a non-signing person can interpret.
Students at the University of Houston designed a device called MyVoice, which uses a video camera to capture a person's sign language movements. It also contains a small video monitor, a microphone and a speaker. Software processes the images and determines what was said, and then translates the word or phrase into speech, which is transmitted through an electronic voice.
It also works backward, capturing a person's spoken words and projecting the appropriate hand sign onto the monitor. Students sampled a database of images to train their software to recognize the hand signs, according to a UH news release. The team used between 200 and 300 images per sign.
It's not clear how well the translation algorithms work — so far, the device was able to translate a single phrase: "Good job, Cougars," congratulating the students who designed it. The team has since graduated, but the team members hope to further development of their prototype and eventually build a functional, marketable device, according to industrial design student Sergio Aleman.
The MyVoice prototype recently won a first-place award at an American Society of Engineering Education conference.
So I'm both a Linguist and an American Sign Language interpreter and I hope you will permit me to shed some light on this. Several algorithms exist for recognizing individual signs mostly because that's the easiest part. Much to the same degree that Kinnect can model sets of gestures.
Unfortunately the scope and complexity of signed languages goes far beyond the individual signs themselves. The problem becomes even more difficult when we consider that the visual/spatial grammar of signed languages cannot be modeled by current methods and individual signs can have upwards of 10 semantically unique variants. The tools used now by linguists for documentation are woefully inadequate for the kind of modeling that is needed.
All hope is not lost though. It's only been in the last 60 years that signed languages have been recognized as actual languages rather than being regarded as primitive gestures.
It's exciting to see people thinking of this though! I predict that within the next few years we will see some breakthroughs is this regard.
my roomate's aunt makes $83/hr on the laptop. She has been without work for 8 months but last month her pay was $8682 just working on the laptop for a few hours. Read more on this site...NuttyRich.com
"I predict that within the next few years we will see some breakthroughs is this regard."
Starting with this, one solid branch off from translation software/device to translation for sign language software/device. I'd say we're seeing the breakthroughs now.
As successful as technology develops towards communication, I feel men and woman will still be confused by the mixed signals that share between themselves, lol.
I had this idea back in 2000 and reviewed this application space while getting my MBA. I had steered away from using camera technology at the time and was focusing on another technology. My conclusion was the market was not big enough to form a company out of and at the time I didn't have any IP to protect. I will say however that with the advent of Kinect, the costs of development are much lower. I could see a larger company taking on this application space as an add on product.
Your right of course. It's insanely difficult even for people with years of experience to correctly interpret the variety of dialects of signs which mean the same thing. It's nearly impossible for a computer to do it quickly as the human brain. Kids learn the intricacies of sign language early and those who learn late--in their 40's and 50's have a very difficult time with it. For the same reason computers are not now even close to be as efficient as a 50 year old who learns it from scratch and so they are miles away from having a breakthrough on this.
How do I know this? Because I'm nearly deaf and lost my hearing slowly and had to pick up sign language at a later stage in my life and it's devilishly hard--but as easy as cake for deaf 1 year olds.
This is a great device! This should be highly publicized and mass marketed!!!