Inconsistent PDF output for Adobe 'read out loud' function

Previous topic - Next topic

eeedelli

Hello everyone.

I've been using Scribus for a while now to create the quarterly magazine for our model railway club. We have a couple of members who are dyslexic, so reading lots of text is difficult for them and one, in particular, relies on his phone/PC to read things out loud as his dyslexia is quite severe. I've been attempting to generate the pdf output in a way that will work for him.

I was previously using Scribus 1.4.8 but found it wasn't creating selectable text in the pdf output. I then found lots of threads about un-ticking the subset box in the font settings and uprating the pdf output to v1.6.

I have now done those (and upgraded to Scribus 1.6.2. (on Win 10) while I was at it.)

Although I can now wipe over and select text in the pdf output, I'm finding that using the Adobe 'read out loud' feature (Files > View > Read Out Loud in Acrobat) is producing very strange results.

Some text reads fine. Other text has massive pauses between words and the effect seems to be completely random. This mirrors what can be seen when you click on the text when in the 'read out loud' mode in Acrobat.

The areas that read correctly are all grouped together in a box. Other lines, however, have no grouping at all and the cursor either lands with no boxing around it or the boxing is a part of a word, or a small group of words. (e.g. the word 'caption' is boxed up as 'captio' then a second box for 'n')

This results in the reader voicing up the words separately, with approx. 1 second silences between them (making it unusable), while the words that are grouped together as part of a sentence sometimes all tumble out sounding as though the spaces between the words are not there, resulting in many mispronunciations in rapid fire, making it largely incomprehensible (though it didn't do that in the examples here).

This seems to be a completely random issue. I have uploaded the .sla file and two slightly different versions of the .pdf output to a dropbox folder here: Example files

As you will see if you download the pdf files, the first one I tried (called 'test') produces audio output for the first couple of lines correctly, fails to read the caption correctly, then reads the final sentence below the photo in three chunks.

The slightly later version (with the longer final sentence - and full file name) fails to read any of it at a sensible speed at all.

It seems that Scribus isn't creating consistent grouping of words and sentences for the pdf output. It doesn't seem to make any difference whether one uses file > export > save as pdf, or file > print to pdf.

Does anyone have a solution to this problem, please?


Nermander

Scribus was mainly designed for producing PDFs for print, this means Scribus does not write "sentences" or even "words" in the PDF, it often places individual glyphs. And that of course makes if very difficult for a program to figure out in what order to read things.

eeedelli

Quote from: Nermander on June 23, 2024, 01:28:11 PMScribus was mainly designed for producing PDFs for print, this means Scribus does not write "sentences" or even "words" in the PDF, it often places individual glyphs. And that of course makes if very difficult for a program to figure out in what order to read things.

True.  Fair point. 

Mind you, these days, it is becoming commonplace for files originally intended for print to be used on (and read aloud by) electronic devices, as a matter of course, let alone for those who might have a disability and really do need that option, who we should not be excluding.

I wonder whether it's something that is possible to fix?  If not, it reduces the usefulness of Scribus in comparison with other products which produce output not only for print but also for devices.

eeedelli

Quote from: eeedelli on June 23, 2024, 02:08:00 PMIf not, it reduces the usefulness of Scribus in comparison with other products which produce output not only for print but also for devices.

Also - how is it possible that other programs that also output to pdf for print purposes manage to do it in such a way that the 'read out loud' option in Adobe Acrobat still works?  If they can do it, it does demonstrate that such things must be possible!

The other point here is that this seems to be a completely random problem - some of the identical sentences in two slightly different versions created from the same original file (as demonstrated in the example files) produce totally different results. That does rather suggest that this is unintentional (or, at least, not specifically considered), which also suggests that something rather odd is happening.  Surely computing output should be consistent and should never produce random results? That would seem very strange.

Nermander

Quote from: eeedelli on June 23, 2024, 03:22:01 PMAlso - how is it possible that other programs that also output to pdf for print purposes manage to do it in such a way that the 'read out loud' option in Adobe Acrobat still works?  If they can do it, it does demonstrate that such things must be possible!

The put the text in PDF "stories" (I think it's called).

Essentially, they place the cursor and then output a full line of text.

Scribus on the other hand may place the glyphs individually. It places the cursor, places one glyph, then places the cursor again, places next glyph. This gives a lot higher precision when placing the glyphs, which is good for print. For print it does not even matter in which order the glyphs are placed, becuase you just see the end result, not how it was "built". Essentially Scribus could first place all a:s, then all b:s etc. You would not see any difference in the end result when printed.

eeedelli

I wonder whether there's some sort of script (javascript?) or something that could be deployed in order to convert the finished Scribus output into suitably-formatted 'stories' that would then work with 'read aloud' but without affecting the visual appearance?

Now that so many people are reading "on-screen" rather than on paper (and those with reading difficulties rely on 'read aloud' as an everyday occurrence), it should be something worthy of consideration, I feel.

The ideal would be for the visual appearance of Scribus to remain unaffected but compatibility with 'read aloud' to be achieved.  I have no idea whether that would be possible but I hope that those who are capable of looking at the situation would give it serious consideration.