logo Association BrailleNet

4th European eAccessibility Forum

eAccessibility of Public Services in Europe

12/04/2010, 9:00 - 18:00
Cité des Sciences et de l'Industrie - Paris, France

  Version française
Printable version

Embedding automated accessible outputs in course production

Gerald SCHMIDT (United Kingdom)
Open University

Speaker's information

photo Gerald SCHMIDT

Since completing my doctorate in English at the University of Cambridge in 2001 Gerald Schmidt has been working in educational publishing, first at ProQuest in Cambridge and then at the Open University in Milton Keynes. Currently the main focus of his work is accessibility in e-production.



[Notice : The short papers of this conference have been prepared by BrailleNet who accept any responsibility for them. But presentation materials provided for download (full-papers or slides) have been provided by the authors themselves]

Figure representing OpenUniversity accessibility workflow : inputs (PDF and VLE) are dropped in a structured content workflow which produce outputs (Daisy talking books, Synthetic, speech MP3, ePub for Sony Reader, Nook, ePub with embedded audio/video, Mobi for Kindle, Microsoft Word documents.

Accessible versions of our course materials have long been an integral part of Open University course production. Approximately a hundred courses are available in audible versions, which have evolved from tape cassettes to CDs to DAISY talking books. Provision is not restricted to students with print disabilities, of course, so subtitles and transcripts are prerequisites for all audio and video clips, for example.

For large population courses, our goal is to record all learning materials in the purpose-built audio recording centre. The audio files are dropped into a structured content workflow which uses a single schema for all course production (print, web, alternative formats). Table descriptions and page numbers (for print items) are added before the talking books are distributed to students.

This paper documents our efforts to achieve as much of this as possible for courses that are not available as high-quality human voice recordings and which often cannot afford the level of hand-crafting that goes into our largest courses. In other words, the aim is to achieve a fully accessible curriculum for all courses including those that lack the economy of scale that makes voice recording possible. Some MA programmes and the materials we publish as open educational resources on OpenLearn may serve as examples.

To achieve this, volunteer human readers are replaced by a synthetic voice and all manual intervention is taken out of the production workflow.

Production process

Our course teams feel most comfortable using Microsoft Word to author materials, so we decided to base our structured content workflow on Word, albeit with some customisation of the user interface.

From Word, we can render to PDF, the virtual learning environment, DAISY and all of our alternative outputs including, ironically, Word, at the push of a button.

If DAISY or another alternative output is chosen, jobs are automatically picked up by a dedicated conversion server.

The server handles metadata, image conversion and resizing, synthetic speech generation, MP3 encoding, ID3 tagging, file compression, and related tasks.



DAISY talking book



The MP3 output takes the form of a single MP3 file with the course title, item title and author as ID3 tags (in keeping with standard MP3 terminology, these are represented as album title, track title and artist).

Tables are unrolled, that is, read out in correct reading order and with guidance concerning the current row and column. Multilingual texts are read with appropriate pronunciation and emphasis.


For ebooks we are targeting two separate kinds of device: on the one hand, the ebooks have to work on monochrome e-ink devices such as the Sony Reader, Barnes & Noble Nook and also (via conversion to Amazon’s Mobi format) Kindle. On the other, we are preparing HTML5-compliant ebooks with audio and video players embedded in the main text flow. We are hoping to push up the level of interactivity available on these devices as suitable hardware becomes available.


This is the format that is requested most frequently by students with print disabilities. Especially when using a screen reader fine-tuned to the needs and learning style of the individual user, a Word document with deeply nested heading levels and the appropriate language mark-up is arguably the most accessible format we can produce at the moment.


All of the tools we use are open source software. Perhaps the most important of these is DAISY Pipeline (automatic DAISY production), but ImageMagick (image conversion and sizing), LAME (MP3 encoding), Ghostscript (EPS handling) and various XML-related libraries are all vital parts of the whole.


The thin glue layer that controls the conversion server – and which is the part that we have written ourselves – is available from SourceForge.net at http://sourceforge.net/projects/daisydemon.


Retour à la liste des articles
Help - Site map - Contact