
Aaron Cannon wrote:
Bowerbird wrote:
it's also worth noting that my "viewer" program has text-to-speech capability. so melissa, you can try it out yourself, and let me know how it works for you. it works just fine for me...
Sorry, but dropping in a couple hooks to a text to speech engine doesn't make a program accessible.
This is true. Text-to-speech (TTS) engines require more than just raw text, as Jared Buck mentioned in a separate reply. TTS prefers understandable document structure, so it can deliver the text in a way that listeners[*] will readily understand. For example, higher-end TTS engines want to know what is a section header, what is a paragraph, and other kinds of structural info. That way the TTS can communicate to the listener, using conventions suitable for the audio realm, the structure of the document. (In the visual realm we use agreed-to conventions based on typographic layout to communicate important document structures -- most people learn the visual cues of document structure without explicitly knowing they are "conventions".) Now Bowerbird's ZML format does provide machine readable structure, at least at the paragraph and header level, and that is commendable -- he just needs to build the interface to the TTS engine which will translate the ZML structure to something the TTS will understand (like "this is a level 2 header to this section".) <smile/> But just dump any raw, unnormalized text into TTS, and it is delivered to the listener in an essentially unstructured way. Even if the TTS engine is programmed to be somewhat intelligent at trying to unravel text document structure, it will oftentimes get it wrong. Yuck. Just spend time with some accessibility advocates at a meeting (I met with George Kerscher and Janina Sajka at a meeting once, and spent a lot of time with George and his guide dog "Nesbit" another time), and one gains a greater appreciation for the needs and requirements for accessibility of texts, and how that affects document formatting issues. Jon [* Listeners not only need to be vision-impaired. There are many uses where sighted people will listen to texts, such as when they are driving or fixing equipment.] [For a cool news item about George Kerscher and his guide dog "Nesbit", see: http://www.guidedogs.com/news-Grads%20F01.html ]