SignSpeak is a three-year project that aims to increase the linguistic understanding of sign languages to eventually create a software that is able to translate signing into text or speech. Similar projects have achieved this for spoken and written language translations, such as Arabic or English.
The project will work on gaining insight about sign language structures from a new point of view: A computer will be trained to understand and translate not only individual signs but whole sentences and conversations. To achieve this the SignSpeak system will track the dominant and non-dominant hands, facial expressions and also body posture. To be able to understand sign language, the computer has to take into account the signs before and after, i.e. the context of the produced sign. So data need to be collected. This is quite time consuming to develop more SignSpeak technology.
This project could have large implications on the Deaf community, possibly allowing them to communicate with hearing people through e-services and e-communication. In a few years a Deaf person might be able to simply sign into the camera and a hearing person could either hear a translation or read it on the screen. It is not yet possible to do so but this project is a step towards realizing new e-technologies for Deaf people. Weather forecast Deutsch GebardenSprache (DGS) at TV studio (prerecorded video) is the first step to get best results.
The EUD is involved in this project to ensure the end-user (i.e. the Deaf community) is adequately represented and is able to use the software. To manage this project, the EUD is working together with four research centres and a telefonica in several European countries: Spain, Netherlands, Germany and Austria.
The project has received funding from the FP7. Project has started from the first day of April 2009. It will be finished in the end of March 2012.
For further information, please go to www.signspeak.eu