From gestures to speech! Portable sensors could translate the language of …

BANNER CECOTEC

It is possible that one day not too far away there will be portable sensors capable of interpreting sign language gestures and translating them into a language like English, which would offer a technological solution to communication problems that arise among deaf people and those who do not understand sign language.

Engineers at Texas A&M University are working on a wearable device that can detect the movement and activity of muscles in a person’s arms. The device tries to identify the gestures that a person makes thanks to the use of two sensors: one that responds to the movement of the wrist and the other to those of the arm muscles. Subsequently, this information is received wirelessly by a program that converts the data to translate it into English.

After a first investigation, the engineers discovered the existence of devices that tried to transform the sign language to pass it to text, but their designs were too basic since the technology they used was based on vision or camera solutions.

According to Roozbeh Jafari, a researcher on the study and associate professor of biomedical engineering at Texas A&M University, these existing designs were not enough, since when a person speaks using sign language, they often make gestures at the same time as performing certain movements in the fingers. That is why the idea of ​​creating a portable device that combined motion sensors and muscle activation sensors was born.

The researchers built a prototype system able to recognize the words that people use most often in your daily conversations. Jafari said that once the team begins to expand the program, engineers will include other words that are less used in order to create a more extensive vocabulary.

A disadvantage of this prototype is that the system must be “trained” to respond to each person using the device. For this training process, the user should be asked to repeat or do each hand gesture a couple of times, which could take at least 30 minutes.

Despite all the work that goes into creating both the team and the system, Jafari is confident that the team will advance more and more during the next phases of development.