WGBH/NCAM Report for DIAGRAM:

Interviews with

Mainstream Publishers and

Publishers of Accessible Instructional Materials

 

April 23, 2013

One of NCAM’s activities during Year 3 of the DIAGRAM Center project has been to conduct interviews with targeted publishers to discover current practices and challenges surrounding image accessibility in order to provide guidance for ongoing DIAGRAM activities. These interviews delved into existing and planned digital publishing processes and are intended to help us better understand the changes necessary for adequately supporting accessible images and graphic content. We hope that by increasing our collective understanding of the process behind accessible content production and dissemination, we can produce better tools and more efficient processes. This knowledge can also contribute to the creation of a strategy for moving from the lab into the marketplace; such a strategy should take into account not only standards, technologies and operational processes, but also business models and political concerns. The in-depth interviews were conducted by NCAM’s Bryan Gould and Larry Goldberg from June 2012 through April 2013, with three mainstream publishers and one publisher of accessible instructional materials. Each organization was interviewed for 30-60 minutes and were provided a list of questions in advance (see Appendix A) although discussions ranged beyond the initial set of questions. The responses from these highly cooperative publishers have already begun to guide DIAGRAM’s work, and many answers echo what we’ve learned from our close working relationships with many publishers, from our advisory board and from attendance at numerous conferences throughout the year. Some of the publishers asked to remain anonymous, and in order to provide the widest possible circulation of these interviews, we have decided to remove any identifying information from this report. Interviews with additional publishers and publishing industry service providers and vendors will take place in the coming year, with results circulated to DIAGRAM stakeholders.

Summary

There were a few major areas of agreement among all of the interview participants:

• Publishers and accessible media producers are dedicated to enhancing the accessibility of their images.
• Widely accepted standards help them make the case internally (with a big shout-out to ePub3, but they wish more platforms would adopt it). MathML is gaining significant traction too. HTML5 and its evolving provisions for media accessibility is next up for widespread adoption.
• The proliferation of image and metadata standards also makes interoperability a challenge.
• Better tools for creating the necessary metadata would be very much appreciated; POET has shown what’s possible but they would like something their authors could use, that can be integrated into their existing tools and processes and that can be readily transportable across platforms.
• All interview subjects would like access to “just-in-time” subject matter experts so that image accessibility can be handled immediately during the production process, with access to accessibility expertise as well. Training of volunteers and production staff (and authors!) would be very welcome.

Accessible Instructional Materials Publisher: Interview #1 – June 11, 2012

1. What is the workflow that brings images into eBooks?

• Audio production volunteers read book in a subject area that they are qualified and trained for. Auditions are conducted by staff.
• It is a  “cold read” unless they have the book at home, volunteers are kind of “winging it.”
• Guidelines now state that image descriptions are moved to the end of the page. So the text of the page is read first. Then the reader reviews the surrounding text to determine context and length of description. All labels and text included in the image is read. Then there’s a tag added to “return to main text.”
• Publisher developed new software which facilitates the image description process. Volunteers are typing in image descriptions via this new tool. Books have some navigation. New books with synchronized text will have image description right where they are in the book with skip-ability. Books are chopped and scanned and OCRed or are also delivered as PDFs. Now volunteers are able to download eBooks from home or wherever to describe images. Volunteers describe images and graphics that need to be described but do not repeat what’s already in the text – description is provided where the image provides additional info. These eBooks either have synchronized digital text with test-to-speech or are synchronized with human narrators.
• The image description tool was new as of mid-2012, so a follow-up interview would be timely and appropriate. Roll out for synchronized text books will be after the summer [of 2012] since there’s still training and recruitment to be done. Since the tool was launched, over 1,000 interested volunteers have anticipated working from home (especially those not near a [production] location).
• This publisher is very interested in, and DIAGRAM provided, an image description training webinar
• With tools to pre-script image descriptions, “…guarantee a poorly read scripted description will be better than anything done on the fly.”

2. Where do you get images for book production?

• From the books themselves.
• [Publisher] does chop and scan and then an outside company marks up a PDF.

3. What metadata is available for those images?

• Text is in DAISY XML and DTB – “prodnote” is used for description.
• DTB does not support “content model” but they are interested in the future.
• [Publisher] is interested in using metadata to create a database of images with description attached. Or even just having metadata trigger a window with description prompts or similar descriptions.

4. What image/production formats are you investing in?

• JPGs and PNGs – SVG would be great but not currently available.
• Many times they receive SVG that contains a bitmap – no good or useful.

5. Where do you think is the ideal place or places to introduce image description in your workflow?

• Already in their production processes – in the studio. With the image descriptions already prepared, then “anyone” can narrate the books with the image description scripts. As long as they pronounce everything correctly.
• Having these prewritten descriptions done as early as possible would be the best.
• Also, each office only has a few content experts whereas the VTEXT tool will provide all of the content experts for every book via online distributed labor, for example, they could connect with a pool of 30 as opposed to 4.

6. What accessibility accommodations are you already making regarding images?

• Data that comes to [publisher] includes a description of the print disability of the reader who requested the book.  So the volunteer reader may read differently for dyslexia than totally blind users. But the default assumption is that the user is blind.
• Due to edition changes nearly every book is a custom job. They do try to reuse materials when possible but this can cause problems with year-to- year editions and regional editions.

7. What are the challenges, where are the most difficult obstacles?

• Getting volunteer readers to not teach the image, not to interpret but simply to read the information and data.
• Consistency throughout the book due to many hours of recording and multiple volunteer readers. Pauses, ums, etc., Bias can leak through.
• Multiple volunteer readers tend to learn from each other as they work through the book and will listen to other readers to glean approaches to image description.

8. What tools would help?

• Peer-to-peer networking, tracking expertise of volunteers and sharing lessons learned.
• Volunteers would be helped if they can hear how TTS will read out their descriptions so that they can be sure that the image description will sound good.
• “Phrase suggester” for images.
• Immediate, on-demand support; sharing of best practices.

9.  What standards would help?

• Standard syntax within disciplines (i.e. dash, hyphen, brace, bracket, etc.,) and terms specific to geology, biology, other disciplines.
• If inhaling image descriptions from other sources, would need it to be compatible. Big question if using a description data base: how do we know that a description is really the same image that you are looking at? Context?
• Interested in Content Model, MathML

10. How do the processes differ between k-12 and higher ed content?

• Same linear process and same tools but readers will be different, especially K-3 versus higher grades. Words are spelled and there are more detailed descriptions for younger readers whereas less so in higher ed.
• High-stakes testing and assessments are a challenge: how to describe without giving away answers.

Other notes

• Sound quality and narrator voice/reading quality is an increasing concern. Many people have good qualifications but don’t read well. Some voice-over training will help.
• How can [publisher’s] operations staff seek help from or ask questions of or have access to the DIAGRAM center? A public forum would be helpful.

Mainstream Publisher – Interview #2 – June 11, 2012

1. What is the workflow that brings images into eBooks?

• .book xml is their format. They do some internal stuff that is proprietary for formats for images.
• Alt text is used for image accessibility – unless the author specifies differently, the alt winds up being the caption information that is automatically captured from the file.
• Using .mobi for Kindle

2. Where do you get images for book production?

• A mix. A lot of screen shots for technical books that authors create and provide and submit (hopefully) in preferred formats. Also staff illustrators use Adobe Illustrator.
• Formats include png, jpeg, stored as .eps with text separate – for vector graphics

3. What metadata is available for those images?

• Publisher’s books are very rich with metadata; they take advantage of what’s there, but could certainly use more.
• Publisher is reliant on authors for description and thinks that production staff is not able to create quality description. Thinks that the POET tool is potentially a good way to add image description.

4. What image/production formats are you investing in?

• They recommend PNGs for screen shots and put in .eps wrapper for call outs, etc. then winds up as a .jpg which is ePub friendly.
• Not a big deal to get metadata from source material (Word or whatever the author is using) and would keep the metadata in the production process all the way to production.

5. Where do you think is the ideal place or places to introduce image description in your workflow?

• Need to rely on authors for image description due to limited resources and to make use of their expertise, but do fall back on their own production staff, and they need ways to make the connection between authors and production staff for descriptions.
• Interested in guidelines and training through webinars.

6. What accessibility accommodations are you already making regarding images?

• Publisher “hand waves” at image description – admittedly a bad practice – telling the author that you “could” do image description if you want and if the author is interested then they do provide guidelines and procedures.
• Providing alt-text for images is part of their authoring guidelines.

• The real need is to incentivize authors to see why they should do image description. Including altruism, reaching a wider audience, helping drive state adoption (when states require accessibility).

7. What are the challenges, where are the most difficult obstacles?

• Relying on author and the production staff as a fall back is not very reliable right now.
• Time and staff are challenges but #1 issue is expertise since production staff do not necessarily have any content knowledge or training on description. So they could do more harm than good.
• Need technical knowledge “on tap.” Images are very, very technical, so they are at the mercy of the authors.
• Authors would want to see image descriptions before they are published, so that would effect the timing of publishing.

8. What tools would help?

• A communication/collaboration tool so a production editor can write quick descriptions and then send a link to the author, the author quickly edits them and embeds them immediately into the publishing tool (deep integration), i.e., a sort of web tool version of POET.
• A Quality Control check would be enabled by this sort of tool.

9.  What standards would help?

• They are dedicated to use of ePub3 and the content model but are awaiting adoption by display software and devices.
• A decision needs to be made about longdesc and/or its replacements.

10. How do the processes differ between k-12 and higher ed content?

• They don’t do k-12

Other notes

• Publisher is very interested in image description and could do this quite quickly; they see this as a goal for themselves.
• Publisher plans to open a database of their images available for search for the public.
• They are VERY interested in a training session on image description.

Mainstream Publisher – Interview #3 – March 7, 2013

1. What is the workflow that brings images into eBooks?

• High resolution images are posted by compositors, and images are gathered from a content management system (CMS) and put into an XML database.
• We use and support output from Quark, InDesign and other tools
• Publisher has a proprietary xml format that is used in-house
• The CMS is also referred to as their “convergent management system,” supporting web resolution as well as high resolution formats
• Often, the image credit is embedded – the photo source – and the file is flattened so the credit is part of image, for legal reasons

2. Where do you get images for book production?

• For their professional development line, images are usually supplied by authors, editors
• For higher ed – procurement is often from stock houses such as Corbis, Newscom and others – they have as many as 20 suppliers

3. What metadata is available for those images?

• Heavy reliance on an older standard: IPTC – International Press Telecommunications Council metadata (like XMP)

– see: http://www.iptc.org/site/Photo_Metadata/ – and http://en.wikipedia.org/wiki/IPTC_Information_Interchange_Model

• Adobe products support IPTC metadata core and extensions (“IPTC Core” is based on Adobe’s technical XMP framework.
• IPTC data includes photo credit, sometimes image description, image number from supplier
• IPTC Core has a field for DESCRIPTION which is defined as: “A textual description, including captions, of the item’s content, particularly used where the object is not text.”
• Here is the help text for tools supporting the IPTC DESCRIPTION field: “Enter a “caption” describing the who, what, and why of what is happening in this image, this might include names of people, and/or their role in the action that is taking place within the image.”
• It would take extra work for them to take advantage of all of the IPTC metadata.
• If they used IPTC DESCRIPTION content for alt-text, it might not be accurate in context
• If photos are supplied by authors, all rules are thrown out – anything or nothing might be supplied
• They have and are updating metadata guidelines for authors – compliance is spotty, technology and STEM authors are good; psychology authors, not so much. Business authors often use Excel which can be helpful.
• Stock houses can be good on supplying metadata – Corbis, Newscom, photo researchers; Getty Images is spotty; others have too many images to keep up with complete metadata

4. What image/production formats are you investing in?

• IPTC and XMP mostly
• They are building into their production process editorial capability for tagging

5. Where do you think is the ideal place or places to introduce image description in your workflow?

• Ideally – the author, then editor; if it is not inserted there, it’s a big question and difficult elsewhere in the process
• Some vendors offer image description for the production phase
• Some vendors are using LaTeX
• One of their production people loves MathML and is trying to get Google to put support for MathML back in Chrome

6. What accessibility accommodations are you already making regarding images?

• alt-text is al they are doing at this point
• IPDF’s ePub3 soon, wish there was more device support to drive adoption, urgency
• Related issue for low-vision readers – they will support RGB workflow soon, instead of just grey scale, which will help for contrast adjustment
• MathML is used for all math text books, and is included in their proprietary XML format: But in their Course Management System (LMS), some platforms prevent proper rendering, so equations are rendered as images
• They use VitalBooksXML – a proprietary XML format
• It takes lots of extra time and QA resources to do MathML but they are dedicated to using it
• They are working on closed captioning of videos, focusing on all newly created ones
• New animations are being provided in HTML5, so accessibility support should be easier

7. What are the challenges, where are the most difficult obstacles?

• Cost
• Writing descriptions – Subject Matter Experts (SMEs) are needed, and they are getting some authors to provide it
• [One division] has some textbooks w/600-700 images
• They are challenged by texts such as an architecture book with hundreds of images
• Educating internal staff about the importance of image description and how it fits into the workflow
• Finding documentation on image description, judging which is best. They have run into lots of disability orgs offering their opinions and they want to know who is the gold standard, who to rely on
• Unified documentation would be great
• Scheduling, fast turnaround make adding descriptions difficult
• The best time to introduce alt-text is during authoring, but the manuscript is often very different from the final product, so changes during production often mean changes to alt-text is needed. Hard to maintain and correct alt-text

8. What tools would help?

• They know about POET and would love to see it integrated with ePub3
• They need a “checker” and QA tools – they can’t ask a production editor to unzip an ePub package, so need a way to look at images and descriptions without opening the entire book file
• A pre-flight tool for ePub
• MathML and ChemML support without need for fallback to legacy solutions
• ChemML for journals (Functional Chemistry) and ChemDraw

9.  What standards would help?

• ePub3, MathML, HTML5
• They have ARIA questions and need info about javascript; in general, better guidelines for industry

10. What about Poet, specifically, the DIAGRAM Content Model, and other DIAGRAM initiatives?

• They saw POET at NFB and at ToC 2012, impressed by it, would like to see it useful outside of present restrictions
• The content model would be brilliant if it was widely supported, and it would be a challenge for them to support. Love the idea of “smart images.”

11. How do the processes differ between k-12 and higher ed content?

• They do not have K-12 content
• But they do have different demands from a variety of products: higher ed and academic, professional learners, Bloomberg books, etc. as well as hundreds of journals

Mainstream Publisher – Interview #4 – March 28, April 22, 2013

1. What is the workflow that brings images into eBooks?

• Images are imported from many formats or created from scratch using the authoring tools.

2. Where do you get images for book production?

• Images can be included directly from the web, from an author’s computer, or created from scratch.

3. What metadata is available for those images?

•  Many annotation types can be added.

4. What image/production formats are you investing in?

• We support as many formats as possible.

5. Where do you think is the ideal place or places to introduce image description in your workflow?

• Image annotations can be added any time before publication.

6. What accessibility accommodations are you already making regarding images?

• Authors can include self-voicing alt text.

7. What are the challenges, where are the most difficult obstacles?

• For images, teaching authors to utilize our accessibility features.

8. What tools would help?

•  Trainings for authors in the creation of meaningful verbal descriptions.

9.  What standards would help?

• A utilized and demonstrated educational curriculum that works for print disabled students!

10. What about Poet, specifically, the DIAGRAM Content Model, and other DIAGRAM initiatives?

• I think it is great and needs a profitable market to motivate publishers to adopt it widely.

11. How do the processes differ between k-12 and higher ed content?

• No difference.

Appendix A: Interview Questions

1. What is the workflow that brings images into ebooks?
2. Where do you get images for book production?
3. What metadata is available for those images?
4. What image/production formats are you investing in?
5. Where do you think is the ideal place or places to introduce image description in your workflow?
6. What accessibility accommodations are you already making regarding images?
7. What are the challenges, where are the most difficult obstacles?
8. What tools would help?
9. What standards would help?
10. Ask specifically about Poet, the content model and get feedback on work that we’ve already done.
11. How do the processes differ between k-12 and higher ed content?