| Partners | Overview | Demo | Progress | Sites | Software | Forum |
In the second year of the eSIGN-project our focus was on
i) content creation in all three partner countries: creating more information sites on the Internet with animated signing;
ii) the further development of tools needed for creating signed content;
iii) further improvement of the signed output of our Avatar, Virtual Guido;
iv) user involvement and continued evaluation of both our tools and the Avatar's comprehensibility.
If you are interested in seeing our public deliverables please click here.
In Germany, the aim was to provide deaf people with signed version of sections of the hamburg.de website. Pages that were translated for the project include:
Click here to see some sample pages.
In the Netherlands, the aim was to help deaf people improve their employment prospects and gain improved access to social services and facilities that are offered by government bodies.
In the UK, deployment provides sign language support for Deaf people on webpages, and signed assistance to help with the completion of online transactions. Four pages from the newly created Deaf Connexions website were selected for translation. Deaf Connexions is a voluntary sector organisation offering support and services to Deaf and hard of hearing people in Norfolk. Click here to see a screenshot of the pages.
An additional UK solution recognises that not all deaf citizens have access to computers and Internet connections in the home, and that not all Government information is available online at present. In the Norfolk area Council Information Centres (CICs) have been provided as a first point of contact for members of the public wanting access to Government information and services. As a result of this, VANESSA (Voice Activated Network Enabled Speech and Sign Assistant) was developed and made available through one of the newly launched CICs. VANESSA aims to help CIC assistants communicate with their deaf clients, by providing Signed assistance for filling in paper forms such as those for Housing Benefit; arranging an interpreter booking for very complex matters; more general issues such as "I will get you a form to fill in" or redirecting clients to more appropriate offices.
The tools for creating signed text have been developed to support both structured content generation and free text.
The Structured Content Generation Tool can build a sequence of signed phrases from a previously created set of phrases. Each output of the tool is a webpage which contains the created phrase sequence signed by an Avatar. The equivalent text is displayed on the same page. For an example of a webpage click here.
The eSIGN Editor software allows the user to compose signed text to be performed by the eSIGN Avatar.
Click here to get more details of the final version of the Editor.
Incremental improvements to the synthetic animation have continued in response to requests from content creators. These have been made available as updates to the animgen.dll and configuration files. Over the report period, eleven updates have been released, the final one being incorporated into the last release of the SiGML Signing Animation software package, v2.7. The improvements include:
The Server Side Rendering system is intended to insure cross platform availability of pre-rendered animated sequences, including the generation of sequences capable of being viewed (as video) on older machines. This is important for eSIGN, as members of the target user groups – individuals within European Deaf communities – cannot be assumed to have access to particularly modern nor even standard PCs with high multi-media capability. This system has been finalised.
We have reviewed the different approaches to content creation that have been used in the course of the eSIGN project. Different tools have been used to produce content for websites and other applications. From this it is possible to define a more general set of workflow requirements for a Content Management System (CMS) for signed content creation. It takes into account the constraints which different content styles make, and considers signed content in the context of the multilingual translation techniques, highlighting the differences particular to sign languages.
Screen shots and example movies of Virtual Guido and his predecessor Visia can be found on our Demo page.
User input has been an important part of our work throughout this project. In all three partner countries, work in close cooperation with both user groups and content providers was continued. A wide range of different types of evaluations were carried out. These have provided information about the linguistic qualities of the signing that has been generated, specifically about how easy the single signs and phrases are to understand. These types of evaluations were carried out with the signing isolated from its intended context. In addition, users have interacted with the finished products. This has enabled people to comment on the comprehensibility of the signing in context, and on the general usability and usefulness of the systems that were developed.
These continued evaluations of the Avatar’s performance as well as of the design of the websites were used for the further development of the tools and the Avatar’s appearance and performance. The evaluations have shown that the Avatar’s comprehensibility has improved considerably over the past months, and that users have a posiive reaction to virtual signing being provided in these applications.