In publica commoda

2.2.5 Transcription

A transcription is used to make the content of the video available in a separate text document. This enables work on the content with or without the video, which can be very useful when preparing for examinations. In order for the transcription to be used by anyone, it should be prepared in accessible Word or PDF format. It can also be provided in several languages.
Available text documents such as the script and/or subtitle file can be used to produce the transcription. The audio description (AD) text must also be supplemented if one has been produced for the video. The AD is inserted as an alternative text (alt text) for the graphics that are used. The information in the alt text can then be accessed as an alternative to the graphics. If there are lengthy descriptions, the alt text can also be integrated in the main text under its own heading after the illustration (e.g. “Description of Diagram X”). Once the text version of the content is finalised, the document is structured hierarchically in line with the video structure and/or presentation (e.g. sections, headings, slide titles). Before publishing, the document must be made accessible so that the information it contains is accessible to all; this is also the case for mathematical formulas. Guides issued by TU Dresden and TH Cologne explain how to prepare Word and PDF files in accessible format. HessenHub’s online self-study course provides a very practical guide. Accessible PDF files can only be produced using paid-for applications such as Adobe Acrobat Pro or AxesWord.
Transcription can include various aspects:

  • Basic transcription is a text version of the spoken and non-spoken content (e.g. text elements of PPT slides).
  • A descriptive transcription also puts into words visual information such as PPT slide graphics that are important to comprehension (via the AD texts)
  • Potentially also translation into other languages.

On the DaLele4All team we have opted for the production of descriptive transcriptions, so that the documents have something to offer everyone. To do this we used the revised subtitle files together with the slide texts and the text versions of the AD. The texts were assembled in Word and structured/placed in a hierarchy along with the presentation slides. We chose to follow the sequence: slide text, screenshot of illustrations relevant to content with alt text and/or formatted table, followed by spoken text. In this way, screen reader users can read the slides first and then have the content of the slides in mind as they hear the spoken text. This is the same logic as the positioning of AD in the video (see 2.2.2 Audio description). Afterwards, the text was formatted in line with the accessibility criteria and exported as an accessible Word document. Provided you know the rules for preparing accessible documents, the transcription can be prepared rapidly.

Example of a transcript structure: The picture shows an excerpt from the transcript of the video 'Frequency tables and histograms'. Under the heading with the slide number 9 follows the slide text together with a formatted table under its own heading. This is followed by the speech text, also under its own heading.
Diagram 10: Example transcription structured and placed in a hierarchy in line with the presentation slides

Realisation involves the following tasks:
  1. Decide on scope (simple or descriptive transcription)
  2. Produce the transcription
  3. Potentially also translation (e.g. English)
  4. Prepare the documents in accessible form
  5. Test with users to ensure that the content is accessible in form

Practical tips

  • Avoid writing mathematical formulas in Word Editor. Many screen readers cannot recognise them, and the garbled output can lead to users being confused. In a test, for example, the JAWS screen reader ambiguously gave a fraction formatted in Word Editor as “x over y”. Therefore, when choosing the output format, make sure that the language of that format is also supported by current screen readers. As it is not possible to equip formulas written in Editor with alt text, formulas are made accessible by inserting a screenshot of the formula in the transcription. The procedure is to incorporate the source code of the LaTeX typesetting system in the alt text for the screenshot. Relevant AD texts can also be stored in alt text (see Diagram 11). In our example of the fraction, the LaTeX code yields clear output as "\frac{numerator}{denominator}". This requires users to understand LaTeX codes. You can for example use Codecogs converter to convert your formulas into LaTeX code.
  • If they have relevant content, tables included in teaching videos require AD that at least gives the subject, column and line titles. In the transcription they are formatted as a table, according to accessibility standards. This makes tables legible by screen readers and it is not necessary to insert the AD texts.
  • If the same diagram is shown several times in the presentation slides, it does not have to be included in the transcription repeatedly. Instead, the repetition of the diagram can be mentioned in the transcription and an in-text reference added to the previous instance, i.e. “see Slide XY” (see Diagram 10). This indicates clearly that the diagram has not been overlooked, simply not repeated. Repetition of the same diagram is however not obtrusive for screen reader users at least, since it does away with navigating back and forth, and it is easy to jump to the next heading with a keyboard shortcut.
  • Diagrams from the presentation can be included in the transcription, even if they are not absolutely relevant to the content. For example, in the teaching video “Daten im 21. Jahrhundert” we included the screenshots of the homepages of Amazon and Tinder in the transcription because they can be enlarged and used more rapidly by students with residual eyesight than image descriptions. However, illustrations like this do require alt text to be accessible to all users.
  • Alt text for illustrations should be brief and succinct. Word’s input screen does not allow formatting. Consequently, it can be worth integrating the AD texts under the illustration with a heading of their own in the main text of the transcription, if the description is lengthy or the text requires specific formatting. If, however, there are numerous lengthy descriptions of images in the main text the transcription can rapidly feel bloated. An accessible PDF can help in this case. In a PDF it is possible to embed a separate file with the lengthy alt text in the main document and to link to the relevant graphic from the alt text. PDFs also allow plenty of creative freedom in the formatting of the alt text.
  • Where possible you should have the finished transcriptions checked by screen reader users. Their feedback on specific details can help with optimal preparation and improve your transcription production process in future.
  • It can also be a good idea to ask screen reader users (if you know any) about their personal preferences concerning the structure and presentation, and take this into account when producing the transcription. Since there are no standard rules for transcriptions, you will repeatedly have to take decisions, based on the logic of the requirements of screen readers and accessible documents. Personal preferences and information can help you to produce suitable transcriptions.

Example of how a mathematical formula can be written accessibly in the transcript: In the text of the transcript, the mathematical formula for forming the sum of the values is circled in color. The formula has been inserted into the transcript as a screenshot. From the encirclement, an arrow of the same color leads to the right to the opened editing window of the graph. The 'Alternative Text' tab is highlighted and selected as well. The image shows the input field for inserting an alternative text for an image in Word. The alternative text starts with the LaTeX code of the mathematical formula followed by the semantic description.
Diagram 11: Example of a mathematical formula that was included in the transcription as a screenshot. The LaTeX code and the AD text were inserted in the alternative text for the image.