Phones for deaf people

Sheila Hemami, a Cornell professor of electrical and computer engineering, led the research with Eve Riskin and Richard Ladner of the University of Washington.

Since their so-called Mobile ASL (American Sign Language) project started four years ago, the researchers have created their first phone prototypes, which are now in the hands of about 25 deaf people in the Seattle area.

The Mobile ASL team designed the bespoke video compression software for the phones specifically with ASL users in mind. Because ASL requires efficient motion capture, the researchers had to make video compression software that could deliver video at about 10 frames per second. They also had to work within the standard wireless 2G network, which only allows transmission of video at about 15-20 kilobits per second.

They also faced constraints such as the battery life of a phone. They solved this problem by writing software smart enough to vary the frames per second based on whether the user is signing or watching the other person sign.

The research also gave the team clues on how people use the phones, said Frank Ciaramello, a graduate student working on the project. They learned that deaf people often use only one hand to sign, depending on the situation, and that they’re very good at adapting as needed. And they found that when two people are talking to each other, they spend almost the entire time focused on the other person’s face.

‘Facial expressions are really important in ASL, because they add a lot of information,’ added Ciaramello. The researchers concluded that their cell phone video would have to be clearest in the face and hands, while they could spare some detail in the torso and in the background. Studies with deaf people who rated different videos on an intelligibility scale helped the researchers hone in on the best areas to focus in their video.

The researchers are now looking for ways to bring down the cost of integrating the software into the phones.