I noticed that there is
a report on github that the rendering of Hindi texts in epub does not work, and that there is no response to this report. A reason for this is maybe that the original report does not explain what exactly does not work.
Since I have experienced the same problem, I can show you on a screen shot what is wrong. On the left side is KOReader, on the right side is nickel (epub). Just for information, the text is Sanskrit, not Hindi, but both languages are nowadays written with the same script (Devanagari). So this should not make a difference.
In this script, there are conjunct consonants. That means two or more consonants-letters melt somhow together to form a new pattern. That two consonats should melt together is indicated by a special sign that is inserted between them.
In this script, there is a special sign, called virāma. In electronic type setting, this sign is used to indicate that the consonant on its left site and the consonant to its right site shell somehow melt together (form a kind of ligature). You can see an example of such a ligature at the line 2. On the right side (nickel) the first character of line 2 is the ligature pr(a), the second character is th(a). Note: the vowel "a" is not actually written, but implied. On left side the first character of line 2 is "p" with a stroke below. The stroke below is the virāma. On this page, you can see several virāmas. In each case, the virāma should not be visible, and instead the letter above be combined with the next letter. For general information, you can for instance have a look at
https://en.wikibooks.org/wiki/Hindi/...t_combinations.
The next problem is the short vowel "i" when it follows on a consonant. The "i" is written as a straight vertical line with a hook to the right on the top. The special thing is that it is on the left side of the consonant to which it belongs.
You can see an example of this on the right page (nickel) in line 4. The line starts with "i", next follows "r", then follows "pu". Together they are read as "ripu." On the left page (koreader), line 5 starts with "r(a)", followed by "i", and then "pu." It is impossible to read this, because you would have to read "ra" and then "p" followed by "i" and "u" simultanously.
I guess it is not in the hand of the developer of KOReader to change this. Or is it?
For those who read Sanskrit: I know that there are mistakes in the text. This is an uncorrected OCR-ed text from a scanned image.