Page break at sentence....?

[se1961]

Here's a weird one:

I have been using and LOVING NaturalReaders Online, to listen to my ebooks. It's a (expensive) miracle! I far prefer it to audio books, which I find over-performed, often to the point of hurting the experience of the text.

The TTS treats all white space as a pause, so when the sentences break across a page, it inserts a weird pause in the middle of the sentence. I am wondering if there is any clever way to get calibre to force page breaks only AFTER a period.

I'm guessing probably not, but miracles do seem to happen in this software--

[kovidgoyal]

calibre does not insert page breaks in the middle of a sentence. It only looks that way because your ebook vieweing software will be paginating the text on the fly. Why your TTS software breaks at the end of these virtual pages I cannot tell you, you will need to ask its developers.

[se1961]

Thank you for this answer. I did understand that somewhere along the way, the pages are breaking “on the fly” since that is how EPUB allows me to increase text size. I import my ebooks into Calibre and convert them there to EPUB, MOBI, PDF, etc. In that process, I am able to give formatting instructions that will force justify, or set margins, or make text a certain size. I guess I am asking, is there some kind of formatting command that would tell the pages to break only AFTER a full stop/period? I am a screenwriter, and there is such a command in my screenwriting software.

[Quoth]

Originally Posted by se1961

Thank you for this answer. I did understand that somewhere along the way, the pages are breaking “on the fly” since that is how EPUB allows me to increase text size. I import my ebooks into Calibre and convert them there to EPUB, MOBI, PDF, etc. In that process, I am able to give formatting instructions that will force justify, or set margins, or make text a certain size. I guess I am asking, is there some kind of formatting command that would tell the pages to break only AFTER a full stop/period? I am a screenwriter, and there is such a command in my screenwriting software.

It's what widows and orphans is for. But your TTS is broken / in error to be pausing between virtual pages caused by text flow rather than file breaks/headings.

However even Widows and Orphan settings won't work because paragraphs larger than a page exist (esp in Victorian & 18th novels.).

I can't see how your screenwriting works? What is the output, paper, screen or Autocue?

I've never seen any such formatting in a Wordprocessor.

[se1961]

Yes, the TTS is broken! I have exchanged a number of emails about the usability problems, which are NUMEROUS. Why do I use it? Because the TTS is breath-taking, often indistinguishable from a real person, reading. Except the TTS currently is engineered to treat any white space as a paragraph break, and that seems to be just baked in.

I have been fussing with this a little bit today, and have discovered that the problem occurs when the software (Natural Reader Online, by the way) imports my ePub-- it automatically converts the ePub into a PDF, and makes a number of bad UI choices-- fonts that look tinier than a list of sunscreen ingredients on my iPhone, and hard page breaks, among others.

I just created a dummy PDF, by taking a long word document and setting it to "keep lines together" as default behavior for the Normal style, setting extremely wide margins and using 20pt as the default text size and then printing to PDF. I discovered to my joy that when I imported this PDF into Natural Reader it was not changed, so the font was big enough to read on my phone, and only VERY occasionally was a paragraph too long for the page.

I am now hoping that if I could get the settings right, I could convert my books straight to PDF in Calibre-- with huge fonts and "keep paragraph lines together" and a page size that would work for an iPhone.... Is such a thing possible? If so, might you have a suggestion about the best margins/font size for phone reading? If not, might you have an idea about another workflow that would do the trick? Could I, for example, convert to html in calibre, import the whole thing into word and do my formatting there...? Or....?

I may sound bonkers, but I would be so grateful to solve this problem. I can't tell you how great the listening experience is, and I LONG for a listening/reading experience that will free me from Kindle/Audible, especially since I can't bear most audiobooks, which are (imho) over-performed....

[se1961]

Providentially, a similar question just popped up in another thread, with a link to the help file that I think will steer me where I need to go! Ill let you know

Originally Posted by kovidgoyal

https://manual.calibre-ebook.com/con...verting-to-pdf

[se1961]

Okay, I have been playing a LOT with conversion and seem to have found a way to convert the EPUB so that I can read it on my phone! Victory! I have one last question:

At the moment I'm converting the epub to DOCX in calibre, then opening the document in word, changing normal style to "keep lines together," and printing as PDF. Is there some kind of CSS command that would force a PDF conversion to keep lines together in a paragraph, so that I don't have to do the extra step? Thanks in advance to any smart person who knows!

PS And/or is there a command that will force the "keep lines together" command when I convert to docx in calibre, so then my only step would be to print the docx as pdf...? Just looking to automate the workflow as much as possible...

[JSWolf]

The only way to keep paragraphs together is to manually set page breaks so the paragraphs do not break.

[JSWolf]

I went to try this TTS and it's broken. It says it accepts ePub and every ePub I tried it gives a conversion error.

[theducks]

A PAGE in an EPUB is a Scroll (chapter file) unless it contains a Perforations (page break). (and AFAIK, pagebreak can not be commanded inside a paragraph)

Devices have screens of limited size, so they either scroll or paginate the file.
TTS should be treating the file like a scroll, so there should be no mid paragraph break.
BUT
If you have some sort of Viewer looking at the file (read out loud), the view MUST control things.

[se1961]

Originally Posted by JSWolf

The only way to keep paragraphs together is to manually set page breaks so the paragraphs do not break.

That is too much, even for me! I guess I will stick with my DOCX workaround. I'm surprised there's no way to set the "keep lines together" formatting command when converting to DOCX in Calibre..

I think from your section post that you were saying you couldn't get "Natural Reader Online" to work with epubs-- it did work for me, when I created the epubs in Calibre. But it was almost useless, because NRO then generates a PDF with completely unusable settings (teeny tiny fonts, etc..) If you want to test it out, I think it make the most sense to create a PDF the way you like it, and then import the file to NRO-- since it doesn't do any conversion to a PDF, I presume it would import without error.

Thanks everyone!

[JSWolf]

Originally Posted by se1961

That is too much, even for me! I guess I will stick with my DOCX workaround. I'm surprised there's no way to set the "keep lines together" formatting command when converting to DOCX in Calibre..

I think from your section post that you were saying you couldn't get "Natural Reader Online" to work with epubs-- it did work for me, when I created the epubs in Calibre. But it was almost useless, because NRO then generates a PDF with completely unusable settings (teeny tiny fonts, etc..) If you want to test it out, I think it make the most sense to create a PDF the way you like it, and then import the file to NRO-- since it doesn't do any conversion to a PDF, I presume it would import without error.

Thanks everyone!

It failed converting the ePub (with no errors) to PDF. And the code is not complicated so there should have been no reason to fail.

[Quoth]

Originally Posted by se1961

I am now hoping that if I could get the settings right, I could convert my books straight to PDF in Calibre-- with huge fonts and "keep paragraph lines together" and a page size that would work for an iPhone.... Is such a thing possible?

LO Writer does perfect PDF exports. Format in LO Writer and export PDF.
Only edit odt after first import to Writer.
Do extra Save As if wanting an epub, Save As docx and import to calibre.

Calibre can export docx or RTF (some ebooks are better exported as RTF) from epubs. LO Writer can import either of those, but save at once as odt for editing and subsequent changes.

[davidfor]

Originally Posted by se1961

That is too much, even for me! I guess I will stick with my DOCX workaround. I'm surprised there's no way to set the "keep lines together" formatting command when converting to DOCX in Calibre..

I think from your section post that you were saying you couldn't get "Natural Reader Online" to work with epubs-- it did work for me, when I created the epubs in Calibre. But it was almost useless, because NRO then generates a PDF with completely unusable settings (teeny tiny fonts, etc..) If you want to test it out, I think it make the most sense to create a PDF the way you like it, and then import the file to NRO-- since it doesn't do any conversion to a PDF, I presume it would import without error.

What that says is that the problem is in "Natural Reader Online". This demonstrates that they have taken a cheap solution, convert to PDF and use someones TTS solution without truly knowing who it works. Have you tried using calibre to convert the epub to PDF and then running the TTS that is on your computer? I suspect you will get the same pause at the end of a page.

But, you have missed something that was sort of suggested. A mention was made of widows and orphans. These are used to keep parts of a paragraph together over the end of a line. The default is 2 or 3. Most people want 1 as that fills the screen as much as possible. Using a much higher number will mean less paragraphs get split over a page. If you use a number that is more than half the number of lines in the longest paragraph, then the paragraphs won't get split (unless they are longer than a page). That has worked for someone on an ereader where they wanted this. I don't know if it will work for the PDF render that the tool is using. If they actually said what backend tools they use, you could get an idea of whether it would work or not.

And for the record, it didn't. And I wasn't impressed with the handling of some of the text. The test case I used has a heading for "Chapter n" and the first letter of the first paragraph after this is a drop-cap. After processing the file, it read that first letter, then the heading and then the rest of the first word. That's pretty crappy behaviour and make me question what else they get wrong.

[se1961]

Originally Posted by davidfor

What that says is that the problem is in "Natural Reader Online". This demonstrates that they have taken a cheap solution, convert to PDF and use someones TTS solution without truly knowing who it works.
—
And for the record, it didn't. And I wasn't impressed with the handling of some of the text. The test case I used has a heading for "Chapter n" and the first letter of the first paragraph after this is a drop-cap. After processing the file, it read that first letter, then the heading and then the rest of the first word. That's pretty crappy behaviour and make me question what else they get wrong.

Voice Dream Reader is a marvelous piece of software that does all the formatting and processing this software doesn’t do, but unfortunately still uses very robotic sounding voices. I reached out to the developer and asked if he would be willing to work with these voices which use “neural processing“. He said there are privacy issues since all the processing is done online. It’s clear that natural reader online is doing exactly what you said: taking advantage of the processing and not spending any time on the software. But actually I have made my peace with it now that I have figured out how to create a readable PDF with no line breaks across pages. There are settings, at least in my phone application, that prevent it from reading headers and other non-narrative text like footnotes.