[Accessibility_sig] speech style sheets?

[clc4tts at HotPOP.com]

  1.) Does Mozilla support such a style sheet?
  Sort of. Technically, Mozilla itself does. However, this support is pointless if screen readers do not support it. Unfortunately, there is virtually no screen reader support at the moment.

  2.) Have any style sheets for speech synthesizers been written?
  This is a two part answer:

  Part 1: As far as speech synthesizers go, they have a lovely way of not agreeing on what a style sheet should look like. Microsoft's SAPI 4 uses a completely different format than its SAPI 5; Sun's Java speech engine uses yet another format. Screen readers, such as Fire Vox, that support multiple speech engines must deal with the problem of taking the same user command and generating different commands for the speech synthesizers depending on what the speech synthesizer expects.

  Part 2: I think you would probably be more interested in screen readers rather than speech synthesizers since speech synthesizers would not have access to CSS stylesheets anyway and regardless of which format the speech synthesizer wanted, it is up to the screen reader to make the proper conversion from CSS style to speech synthesizer controls. The CSS styles that are for speech are aural style sheets in CSS 2 and the Speech module of CSS 3. However these two also have their differences. And then there is the pesky problem of getting the CSS information from the browser to the screen reader in the first place.

  3.) Can they be applied (through Mozilla) to the websites a browser visits?
  Mozilla does allow user style sheets, so technically, yes they can be applied. However, as I mentioned earlier, they are not too useful because of a lack of screen reader support. JAWS does not support aural style sheets and I do not see any indications that they plan to in the near future; I do have plans for implementing it in Fire Vox, the freeware, open source screen reader that I develop, but that feature is not ready yet. Opera apparently does have some support for the CSS 3 speech module.

  I'd also like to point out some of the current difficulties with supporting such style sheets:

  1. As a developer, I have to choose between the older and somewhat better known CSS 2 aural style sheet, or the newer but lesser known CSS 3 speech module (and possibly having to do both). Having two standards is painful.

  2. One of the rather disturbing things that I have seen suggested with CSS style sheets is that they can be used to add additional meta-information about the content on a page. For example, I have seen suggestions that say these style sheets should be used to precede all headings with the label "Heading Level X" when read by a screen reader. I am sure these suggestions were made with the best of intentions; nonetheless, this idea is a mistake in my opinion. Screen readers such as JAWS and Fire Vox are already automatically applying this information to pages as they are read. As I look more and more into adding CSS support, one of the major problems that I am wrestling with is how to balance between announcing meta information and using content from style sheets. If I use style sheets all the time, then users will actually hear the Heading Level X announcement twice. And while I do plan to let users turn off my Heading Level X announcement, it means that if they were to reach a page that has no such CSS style sheet, they will have to turn my announcements back on, which is at best annoying and at worst unusably confusing. 

  A good way to resolve this issue would be if there were some tag for CSS to identify the rationale behind the style: for example, whether the style provides element identification information or additional new information that is not obvious from the element type (such as how to pronounce the word, or perhaps some important aspect of the visual layout of the text that would be lost to visually impaired users who are simply listening to it). I could then have a rule for my program that would make it read everything that is additional information; but if it came to element identification then users could toggle between my announcements or the CSS style provided by the web page author. And if there were no element identification defined in the CSS, it would always default to my identification announcement. However, as far as I know, there is no such rationale tag in either CSS standard.

  3. There is a dearth of resources for screen reader developers who seriously wants to look into this problem. I have been googling for hours trying to find CSS aural and speech module style sheet examples; what I have found are snippets of code, but what I want (and desperately need) are full, complete style sheets and test pages. I have a limited amount of time to work on Fire Vox each day, and given the upcoming Firefox 1.1 release and the testing needed to ensure that Fire Vox will be compatible once 1.1 is released, I simply cannot afford to spend much time in trying to build style sheet test cases from scratch. This lack of complete examples is greatly hindering my progress.

  I hope the above information has been helpful. 

  -Charles L. Chen
  Creator of Fire Vox, the freeware, open source screen reading plug-in for Mozilla Firefox
  http://clc-4-tts.cjb.net
-------------- next part --------------
An HTML attachment was scrubbed...
URL: /pipermail/accessibility_sig_knowbility.org/attachments/20050705/a5740676/attachment.htm