If by "Acrobat reader" you mean the "read out loud" feature - that is not a suitable tool for checking the accessibility of tagged PDFs. You should try more suitable tools instead, such as
- NVDA (free screenreader, Windows only, see NV Access site at https://www.nvaccess.org
- VIP Reader (free reader for visually impaired persons, Windows, Mac and Linux, for a good intro see Access for all site at http://www.access-for-all.ch/en/201-englisch/500-vip-pdf-reader-e.html and for download go to Swiss Central Organization for the Blind site at https://www.szb.ch/en/hilfsmittel/digital-tools/the-first-pdf-reader-for-visually-impaired -people/ (at the bottom of the page)
- PAC 2 / PDF Accessibility Checker, specifically the "Screenreader Preview feature (free, Windows only), see Acess for all site at http://www.access-for-all.ch/en/pdf-lab/455-new-test-pdf-files-with-the-pdf-accessibility- checker-pac.html
Of the two assistive technogloies VIP Reader and NVDA, VIP Reader (in comparison with the NVDA screen reader) is very easy to use for non-disabled people.
FWIW, there is no such thing as “EPS in PDF.”
When you place an EPS file into InDesign, InDesign totally internalizes the PostScript content into the InDesign document and such EPS content is indistinguishable from other InDesign content by the time you have an exported PDF file.
You may wish to also post this in the InDesign forum for the perspective of other InDesign users with similar issues.
Yes, text to speech reader is how I meant that. We have been using PAC2 to test and troubleshoot errors. We just happened to open the PDF in Acrobat to do a quick read through when we noticed this issue.
Thank you for an extended list of test/read options.
I wanted to provide as much context as possible to help troubleshoot this by mentioning the artwork file type used in the InDesign side, in case that affects the PDF or at least the perspective of anyone here considering replying with ideas.
I had hoped that the grouped object would be recognized as a group rather than individual text elements within the grouped element. While I am not describing that technically accurate, I hope it conveys my meaning!
Thank you for your clarification and for the suggestion to post over in IND. I will cross-post there as well.
In context of your response, do you know of a way to group/tag/whatever PDF content (for example, a small bar chart on a page) that would 1) prevent the text-to-speech reader from reading the internal content of that group, and 2) still allow Alt Text to be read in that group's place?
Thank you again for any insight.
1 person found this helpful
I don't see anyway to do this automatically from Indesign. You would have to manually tag the document and artifact ONLY the text to get Adobe Reader to ignore it. Alternatively you could create an invisible dummy graphic behind the real graphic and put the alt text on that. Then artifact the real graphic. This seems like a lot of work to get Acrobat ReadOutloud to act like a true screen reader.
When I reproduced a simplified version of your problem by creating a Word file with one sentence and a box below it, creating and then manually tagging a PDF, this is what I found.
1. tagged all content as a figure with alt text. Adobe Read Out loud reads both the alt text and the tag contents in XI, DC and DC Reader.
2. added "Actual Text" to the Figure tag, same result as 1.
3. tagged the sentence inside a Span tag inside the Figure tag, same result as 1.
4. added Alt text to the Span tag and Read Out Loud reads the alt text for both but does NOT read the content. You could conceivably define null text in the Span tag and this would eliminate the problem. This is not recommended and it causes a "nested alt text error when running Acrobat Accessibility check in XI.
5. artifacted the text that was in the Span tag and deleted the empty Span tag. It now reads correctly and no error generated in Acrobat XI, DC, and DC Reader (although Reader took a few restarts, at first it wasn't reading anything).
I found no problem reading the original document tagged as in #1 above with either Jaws or NVDA screen readers using Acrobat XI or Acrobat DC. They are legitimate screen readers, Assistive Technology -- they are screen readers. Read OutLoud is "Text to Speech" tool according to Acrobat.
Why do you have to retain object-oriented content (i.e., EPS content) in your final document? This problem is easily resolved by converting all graphics to bitmaps and some U.S. agencies still prefer this because the document is more stable. In addition to resolving the reading issue you will not have objects floating under or over other objects on a page once they are bitmapped.
My apologies for the late response. Thanks for the detailed reply btw! I think most of our issue is that this is simply new to us so we're making very rookie mistakes. Testing in NVDA worked fine for the most part for, as you noted, that is an actual screen reader where Acrobat is simply Adobe's text-to-speech. Our bad there for starting a fire that might not have really been as urgent as we assumed.
Regarding your question as to retaining the layered EPS/AI format, that is mostly a work flow efficiency. Most of our InDesign work is based on frequently updated tech manuals that use layered Illustrator files as visual references. We once tested a book development in which we converted our vector files to raster, and while that may be a path to solution, this process created more headaches farther down the development line. In that our books are regularly updated, we typically have to modify many of the illustrations on the fly. If the illustration has been converted to raster, we then need to pull the original AI, edit, save, export to JPG (or whatever format), to overwrite the previous export in layout. That was a nightmare keeping hundreds of illustrations cross-referenced to hundreds of raster conversions for the inevitable updates that would come.
Perhaps this is a workflow issue that I am not seeing then? We have adapted our workflow to what our primary work efforts demand as far as speed in development. That workflow may not be applicable in 508 compliance. I really don't know. Ideally, I would like to find a compromise between what we must do for current project demands and what we must adjust to for 508 work. I'm not sure what that is yet though.
Do you literally convert AI to raster or set all text to outlines in your projects? If so, how then do you personally handle edits for document version updates?
Thanks again for the detailed response. Much appreciated!
I've come across this issue earlier today and it's frustrating (in my case, actual text in a frame that looks like a pie chart due to the font being used).
I ended up marking the chart as an artifact and added a 1-pixel transparent image in a frame situated on top of the original image that I've tagged instead. This allows me to update the underlying layer without worrying about converting it.
So far it seems to be working out but I have submitted the document for review to the Washington Talking Book & Braille Library to see if they can suggest improvements over the process. Will report back once I hear from them.