[INDOLOGY] Search for Devanagari Fonts

[Harry Spier]

Dear Jan,
You wrote:

> Actually, for training an OCR, it doesn’t matter whether the fonts are
> Unicode or not,  . . .. It needs a bit of extra work to map the shapes to
> the correct classes,
>

My experience with non-unicode fonts  to construct devanagari text (many
fonts over many years) is that it is extremely time consuming.

Its not that there will be one codepoint for one letter.  Ligatures will
have their own separate codepoint and vowels will have one codepoint for
word initial vowels and another for word internal vowels and some letters,
especially metrically long syllables will be constructed from 2 or more
codepoints. So typing sanskrit text with non-unicode fonts  is more like
typesetting than just typing text in. This is especially true with older
non-unicode fonts which will be 8 bit fonts (only 256 codepoints). And each
non-unicode font will use its own mapping so you have to set up your
keyboard (or make a chart to use) for each individual non-unicode font.

Thanks,
Harry Spier

>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://list.indology.info/pipermail/indology/attachments/20250510/4bd5c5c3/attachment.htm>