Abstract
SVG includes features that help to make SVG graphical information accessible to people who are blind or dyslexic. ViewPlus’ IVEO technology provides SVG authoring and viewing applications that make this promise a reality. ViewPlus is now collaborating with the DAISY (Digitally Accessible Information SYstem) consortium to develop author guidelines and to expand SVG to provide features making DAISY SVG excellent both for making current graphical information accessible and for making future mainstream information more useful and more fully accessible.
Table of Contents
SVG 1.0 and 2.0 include two features, title and description properties for graphical objects, intended to promote accessibility of SVG to people with print disabilities. ViewPlus has exploited these two fields in its IVEO technology to permit SVG images to be accessible to blind and dyslexic readers [1], [2]
The ViewPlus IVEO Viewer is used along with a tactile copy of the image to allow Audio-Touch [3], [4] access to graphical information. The blind reader typically has a tactile copy (made by printing the SVG page from IVEO Viewer to a ViewPlus embosser) mounted on a touchpad When text or a graphical object is tapped, it is selected and the text span or object title respectively is spoken by the computer. An object description is spoken when the user presses a computer hot key or double taps the object. The IVEO Audio-touch method provides excellent access by people with print disabilities if the SVG objects and text spans are organized into semantically-meaningful structures and important graphical objects provided with semantically-meaningful titles and descriptions.
IVEO Creator is a WYSIWYG SVG authoring application that can be used to create simple SVG images and add labels to the graphical objects. It has limited authoring capability. It is intended primarily to import SVG and add titles and descriptions to the SVG file and to important graphical objects in the image. It can also add other features such as human-recorded information. Human speech is more understandable than what is normally provided by speech engines.
IVEO Creator Pro extends those capabilities by permitting files in any electronic image formats to be imported and converted to SVG. In general, vector graphic structure is preserved, and text converted from its original format to SVG format. Bit map images can also be imported. OCR (Optical Character Recognition) of bit map images interprets bit map text and overlays it with text in SVG format. However there are no graphical objects in bit map images, and OCR conversion errors can affect text. In principle, vector files provide not only better visual images but much better access to people with print disabilities. This promise is seldom realized in practice however.
Unfortunately, most common graphical authoring applications do not export well-structured SVG. The lowest level graphical objects are often semantically meaningful but are seldom organized into any meaningful structure. Text is seldom organized into semantically-meaningful spans. Many graphics authoring applications produce SVG and other format files with text words broken into several spans or with text expressed in local font tables that are often difficult to convert to unicode. When such text spans are spoken by a speech engine, the output is difficult or impossible to understand.
Many practical difficulties affect text in scientific diagrams. Scientific graph authoring applications often put several semantically different text strings into a single span. X axis labels for example are often in a single span. Touching any x axis label causes all labels to be read aloud. Scientific graphics frequently have labels with Greek and other scientific characters. They often have sub- and super-scripts and occasionally relatively complex mathematical, chemical, or other specialized expressions. SVG does not anticipate complex expressions, and it is not possible to express many math equations in a single semantically-meaningful SVG text span. Even when SVG structures are available for such expressions, authoring applications do not export the equation in that structure. Greek characters, bold or other emphasized characters are often placed in text spans separate from that of the full expression. Commonly all bold characters in an image may be contained in the same text span even though they may belong to entirely separate scientific expressions. These authoring conventions can produce good visual text but be semantic nonsense.
The DAISY (Digital Accessible Information SYstem http://www.daisy.org) consortium has taken a leadership role in trying to overcome the practical difficulties discussed above. DAISY is an international consortium of libraries and other agencies serving needs of the blind and dyslexic. The DAISY logo is widely recognized as the international information accessibility brand. The DAISY XML specification is widely used by agencies around the globe to make information accessible to people with print disabilities. It is being integrated with the epub electronic book format and is increasingly being recognized as an excellent mainstream electronic publication format.
DAISY has formed an SVG Working Group to develop authring guidelines and to expand SVG to provide a good format for excellent accessibility to graphical information. Several ViewPlus developers are leaders of that DAISY group, and IVEO is the vehicle for testing and implementing DAISY SVG. This work is still in progress, so this paper may be considered an interim report on progress.
The DAISY SVG Working Group is presently testing a number of optional additional structures and attributes. Some of these are intended for use by agencies and publishers making materials designed for people with print disabilities. Many have solid mainstream uses as well.
DAISY layers are structural elements permitting text and graphical objects to be grouped to have a broad set of useful properties
The set of DAISY layers are defined by a sequence of form
<daisy:layers>
<daisy:layerItem layer_attributes />
<daisy:layerItem layer_attributes />
...
</daisy:layers>
The layer_attributes control properties of graphical elements assigned to that layer. <layerItem> may have the following attributes:
Any SVG visual element may be assigned to a layer by giving it the attribute daisy:layer = "layerID".
A DAISY View permits one document to show many different portions of the image (eg a full map of the US as well as a view of northeastern states and a view of Washington DC).
The set of views is defined by the element
Each viewItem has the following attributes:
x, y, width and height attributes are given relative to viewBox. Default view for example, equivalent to the full svg viewBox, is x = "0." y = "0." width="1." height="1.".
DAISY Actions provide a flexible alternative to the <title> and <desc> children of SVG graphical objects as well as additional possibilities for linking to XML and multimedia information. DAISY Actions <actions> are additional children of the link element <a> attached to graphical objects. They provide an arbitrary number of text elements instead of the two provided in standard SVG. They also provide for an arbitrary number of links to multimedia information. Currently <title> and <desc> are ignored by IVEO Viewer if the <actions> element is present. <actionItem> child elements are accessed sequentially.
An action item can be text or XML to be spoken by a speech engine or a multimedia link (eg a human voice recording) to be played. <actionItem> elements have attributes: type, value, param.
There currently are two types of actions: type="speak" will speak the text or XML specified in the action value. Type="link" will execute the link specified in action's value. Action param is an arbitrary optional parameter. Currently it is used only to specify the volume when the action is a link to a recording.
SVG documents do not always have semantically well-defined graphical objects. Bit maps for example are a single graphical element, even though there are typically a number of semantically meaningful visual elements in the image. SVG permits use of invisible overlays to define text and semantically meaningful graphical objects in such cases. Invisible overlays are used extensively in IVEO documents, but they are clumsy to manipulate and require the end user to set display options correctly for these overlays to remain invisible. DAISY overlay is an element attribute that provides the author more flexibility and does not pose a burden on the end user. IVEO Viewer makes elements with this attribute invisible. They can be visible for editing in IVEO Creator and Creator Pro. These elements are normally active, providing users the expected information about the text or object it overlays.
The DAISY text structure specifications are being developed as a joint project of DAISY, ViewPlus, and the Infty group http://www.inftyproject.org. The Infty group has developed math OCR capability and is working to modify its Infty reader math OCR application to recognize text and math in SVG files. It then organizes them into the appropriate DAISY text structures. This capability will be licensed to ViewPlus for its IVEO Creator Pro application.
The authors are grateful to the US National Eye Institute of the National Institutes of health for Small Business Innovation Phase I and Phase II grants that have supported much of the SVG developments.
[2] Directly Accessible Mainstream Graphical Information. Lecture Notes in Computer Science 3118(2004) Computers Helping People with Special Needs: 9th International Conference, ICCHP 2004 Paris, France, July 7-9, 2004 Proceedings, pp. 739-744 .
[3] Nomad: an Audio-Tactile Tool for the Acquisition, Use and Management of Spatially Distributed Information by Partially Sighted and Blind Persons, Proceedings of the Second International Symposium on Maps and Graphics for Visually Handicapped People, King's College, University of London, pp. 24-29 1988.