There was a point where just being able to read a book at all was considered wonderful enough and better scanning quality was the best we could hope for. I'm appreciating all the thought it seems is now being put into formatting and other improvements. That being said, we're really working with very inadequate tools in trying to do better with this. We're asking the question as to the best compromise solution considering that there are probably a half dozen different formats in which people are reading these books - braille with several different page widths, text-to-speech with some people just sitting back and listening and others actively moving through the text as they read, some with a screen reader on their computer and some with a DAISY player on a computer or stand-alone device. Some use enlarged print where some enlarge the text meaning again we have different page widths, and others leave the original text and use software to magnify their screen image. Some are using combinations both looking at the words and listening. I've occasionally had things I did with both speech and braille, although I don't know if anyone actually reads that way. There are probably more that I'm not thinking about.
It's impossible to find one best way for all of these. The answer is in being able to use formatting and style codes, or at least in being able to standardize. and then for the final formats to make use of those codes. So if you code something as a page number, when converting to braille, it can be, for example, placed in the top right corner regardless of how wide the page is, could be spaced differently for different presentations of enlargement, and could contain a code that lets the daisy player actually know it's a page number. a line of dots in tables of contents could be present in visual and braille presentations, adjusted for page width, and be active links to the actual page.
I'd thought bookshare was doing some of that, but as I do tend to ignore formatting when I read, I can't say I've noticed. I don't use daisy players or anything else fancy as I don't like the speech engines they use. I'm sure they must certainly be doing this with the NIMAC books that we aren't allowed to access. I was going to go poking around in some of the books I already have, but it would be easier for someone who already knows to give an answer. Even though they don't ask volunteers to add in codes, I'd assumed they did some things by automation, like coding as page numbers sequential numbers that appear at the beginning and end of pages. If they're not, considering all the work volunteers are now putting into these books, it seems we should ask them for a few codes we can use. Validators who chose could then properly code tables of contents, chapter headings, page numbers, and footnotes, at the least. The volunteer manual I saw did recommend enough standardization of such things that it does seem bookshare could be making use of such efforts in the conversion process. Maybe they're afraid not enough people would validate books if they were expected to do this, but since some *are* doing it, maybe we could get some guidance from them. Maybe if they are not making full use of our efforts, we could prod them?
Am I correct in my new reckoning that there is a gap between volunteers and paid staff, that people making decisions about what to automate and how to convert books are not interacting with people doing scanning and validating? Is it that the hopes, which I share, are pinned on getting text from publishers in the future and thus not needing to go through all this? Well, even then we'll still need these tools to include older books in the collection. In the past validators were asked to do simpler things like make sure all the pages seem to be there. It's a lot more now and I think that's good, but what a shame for us here making such compromise decisions when we could do something that will really be used properly. Has this been discussed already?