Web Posted on: January 7, 1999

A METHOD FOR ADAPTING SCREEN PHONE SERVICES FOR VISUALLY HANDICAPPED USERS

Dominique BURGER
Marine FERRÉ BLANCHARD
Djamel HADJADJ
INSERM U483 - Universités Pierre et Marie Curie

Introduction

For years the telephone has been a media naturally accessible to the visually handicapped people who use it as a simple way to communicate with other people, to access information and to overcome some of the difficulties associated with their handicap. But the emergence of phone services requiring screen for more complex interaction with the end user, also called screen phones, not only creates a new barrier for visually handicapped accessing traditional services but also might exclude them from the new information society where telephony will be an important medium for accessing information and a great variety of services based or not on the Internet. [1][2]

The Vistel Consortium has been set up with the support of the EC to explore solutions to this problem and to demonstrate them [3]. In this paper we focus on the technical strategy followed in the project, presenting the general model of interaction and the adaptation technique that were used to adapt an existing screen phone.

The Vistel Problem

• The services considered in Vistel are mainly aimed at delivering textual information and are based on textual interaction methods. Translating images or figures is out of the scope of Vistel.
• The screen layout of the services considered in Vistel are assumed to be stable for a given service..
• Occasional users and users without a long experience of computers should be able to access phone services after only a short and simple training. Thus, the solutions proposed in screen readers currently on the market are far too sophisticated..
• Compatibility was sought with a great variety of standards as well as with the evolution of services in the near future particularly with services delivered over the Internet ..
• The adaptation method shall support either Braille, speech or Screen Magnification.

A General Model of Interactions

A review of a great variety of phone services made it possible to define a general model of interaction covering most of the interaction situations encountered. This model is based on five main principles:

1. Standard functional components,
2. Sequentialisation,
3. Specific Guidance,
4. Non verbal feed back, and
5. Differentiated actions on Keys

Principle 1 : Standard functional components

A screen can be modelised using a five basic Interaction Components providing mechanisms to explore the content of the component and interact with it.:

• texts : purely passive information chunks ( For instance a title, a section of an informative text, a warning message are texts.)
• buttons: used to activate commands
• input buffers: used for typing information in
• menus: which present the user a possibility to make a choice among several items
• Soft Keys: which are programmable keys, defined by the service.

Thus their meaning according to the context changes and their labels are refreshed on the screen. Soft Keys are problematic for visually handicapped users using speech adaptation, as they can not rely on a stable layout of the keys and would therefore easily perform wrong operations.

Nevertheless, all these components are quite traditional and non visual equivalents have been developed for the need of the visually handicapped users, in the context of MS-DOS or Windows, based on speech, Braille or magnified presentation of data.

Principle 2 : Sequentialisation

The presentation of data on a Braille display, or using a speech synthesiser, or magnifying the data on the screen, has to put the components in a sequence. The order of the components in the sequence will be chosen according to the context of the application and can be different for users who do not have the global view of the screen. Assuming that in the context of screen phone services the number of components in a sequence is limited, it is quite realistic to provide a very simple navigation mechanism based on two main function "Go a step forward" and " Go a step backward"'.

Principle 3 : Specific Guidance

Some extra guidance has to be provided to compensate the absence of a global view of the screen. This guidance can be given using textual comments just before an Interaction Component or just after.

Principle 4 : Non verbal feedback

Non verbal feedback is extremely useful in non visual interfaces in order to convey feedback rapidly. It can concern events that the user is expecting and wants confirmation about - like going to the next or previous line or element - or to give low level warning like "no more element after this one". Using beeps is a simple and low cost method to achieve this.

Principle 5 : Differenciated actions on Keys

Visually handicapped users may take advantage of the several possibility of pressing a key in order to get additional information before activating a command. This is particularly useful to solve the Soft Key problem. Pressing a Soft Key once gives information on the associated command while pressing it twice would activates the command.

It is worth noting here that this model is quite general and that it can be applied in quite different contexts. For instance it is fully compatible with the basic HTML elements and interaction components. Also this adaptation model implies simplification, reformulation, reorganisation of the data, as well as modification of the interaction principles used in the visual interface. This is what is called screen reconstruction.

Adaptation technique

The functions of the communication module are :

• To transfer the information displayed on the screen to the adaptation box as soon as this information appears;
• To inform the adaptation box when a full screen has been completed;
• To emulate any of the actions usually performed on the phone device to control the dialogue with the service.
• Reconstruction Module: The services considered in Vistel are based on series of predefined screens whose layout is stable enough over time. It is therefore possible to prepare a reconstructed view - or profile - of each screen - according to the Model of Interaction.. The reconstruction process is done manually and the corresponding profiles are stored in a file for later use when the service is accessed.

(Fig. 2)

This method makes possible that

1. the reconstructed screens present the elements in the order that have been defined, possibly different from the initial order,
2. part of the information displayed on screen can be ignored if it is estimated not to be useful.

The Non Visual Interface presents the data of the reconstructed screens to the user, using either a Braille display, a speech synthesiser or enlargement of text on the screen. It also analyses the actions performed by the user on the available input devices and transmit the corresponding event to the screen phone. This module also manages non verbal feedback.

Vistel demonstration

For the Vistel demonstration the strategy we have described has been applied to adapt an existing screen phone (Philips P100). The architecture described above was implemented in two separate devices, the communication module being developed as an extension of the P100 software while the reconstruction and the non visual interface are implemented in an external device called the Adaptation Box. The Adaptation Box main component is a 486 microprocessor running a MS-DOS operating system. The P100 and the Adaptation Box communicate by means a serial communication port. The Adaptation Box has a secondary serial port and a parallel port to communicate with speech/ Braille devices. Thus the modification of the P100 is only a minimal software modification that can be downloaded or even run on a removable Memory Card. Three main software programmes have been developed to be run by the Adaptation Box

The Screen Profile Editor Software (SPES) has been developed to carry on adaptation of the services manually. It makes it possible to go through a service and for each screen to define Identification Keys (part of the data displayed) and to specify how the screen should be presented through a non visual interface. The Identification Keys and adapted screens are stored in files corresponding to the services.

The Screen Adaptation Software (SAS)

Once the Screen Identifiers and the Screen Elements have been defined, this programme performs dynamically the identification of screens sent by the services and their non visual adaptation. Conversely, it transforms the elements of the rebuilt screen data coherent with the original screen. SAS also generates non verbal feedback using different types of beeps.

The Set Up Programme (SUP) makes it possible to adapt the software to the various hardware configurations used in the Vistel project (Four Braille bars, four speech synthesisers).

Discussion - Conclusion

A General Interaction Model and techniques to adapt Screen Phones have been defined in Vistel. A prototype demonstrates them on various sample services (P100 Internal Services, National Phone Directory accessible via the Teletel Network in France, a ADSI service for purchasing CD, and a Web Service). In all this cases the method proved to be flexible, making possible to reconstruct the visual interface to produce an easy to use non visual interface. This experience conducted the project team to formulate general recommendations that will be made available to three categories of professionals, namely screen phone manufacturers, service providers , and professional users having in charge the adaptation work.

The adaptability of the architecture and reconstruction methods increases the independence between the service and the adapted interface. Thus it minimises the constraints on both screen phones and devices providers and makes the recommendations minimal. Finally the Vistel model can easily be applied in the context of Web services as it is fully compatible with the definitions of HTML elements. Moreover, it shall be underlined that - in the context of Internet - the adaptation work is alleviated since the functional components are specified in HTML so that the reconstruction process can be made automatically.

References

[1] VISTEL Deliverable 4.3-A, Lesson learned from the Minitel experience in France and their influence on the Vistel solutions, June 1998

[2] Hadjadj D., Bouzidi A., Burger D., An HTML Interface for visually handicapped people, 3rd TIDE Congress : improving the Quality of life the European Citizen, IOS Press, 1998, pp. 38-41

[3] http://paperino.cselt.it/ufds/vistel/technappr.html

[4] Burger D., Improved Access for the visually Handicapped : New Prospects and Principles, In : IEEE transaction on rehab Engin., 1994, pp. 111-118

1 Italy : CSELT, Telecom Italia, ORIGIN Italia, UIC ; France : France Télécom/CNET, Universités Pierre et Marie Curie / INSERM-CREARE, AVH, ANPEA ; Pologne : Technical University of Wroclaw ; United Kingdom : RNIB, University of Hertfordshire ; Spain : ONCE ; The Netherlands : Origin BV / Philips Home Services International. Vistel consortium : Italy : CSELT, Telecom Italia, ORIGIN Italia, UIC ; France : France Télécom/CNET, Universités Pierre et Marie Curie / INSERM-CREARE, AVH, ANPEA ; Pologne : Technical University of Wroclaw ; United Kingdom : RNIB, University of Hertfordshire ; Spain : ONCE ; The Netherlands : Origin BV / Philips Home Services International.