Optilon wrote: ↑Tue Sep 15, 2020 7:34 pm No, this is the Roman IME thread. For Devanagiri see: viewtopic.php?f=13&t=14

Ah, sorry.

I am just curious how were you able to represent the 44 devanagari letters with 26 lowercase roman letters. Did you encode effect of shift separately? If yes, then why does it not appear in the frequency graph? Is it not accounted for? Also, I hope you assigned a unique character to each IAST letter...

Optilon wrote: ↑Sat Sep 12, 2020 9:11 am @hurrdudd
Which one is closest to the way one would type?

For your use (optimizing Hindi keyboard input) both Harvard-Kyoto and IAST would work as they assign a single character to each keystroke. As I mentioned elsewhere, Roman readable does not do that.

Isn't a symbol layer a feature that especially programmers would like to have? :D Is it hiding the symbols behind a keystroke? Imagine typing the following with such layout void greet(Person const& person) { cout << "Hello" << person.name[first] << "..."; } The constant switching will very quickly ...

@hurrdudd I now understand why you said it is a bad idea to put diacritics to a third or fourth layer. 13 of the 54 most frequently used symbols are diacritics. I'm surprised the keyboard optimizer was able to tear the text corpora apart and even noticed the diacritics separately. This is the frequ...

@hurrdudd Sorry for asking so many questions. Thank you very much for sharing all your knowledge with me so far. On the wikipedia site for the devanagiri standard layout, and extended devanagiri layout by the government is mentioned. Is it already available somewhere? https://hi.wikipedia.org/wiki/...

Optilon wrote: ↑Thu Sep 10, 2020 4:28 pm Do you have a more detailed picture of the BolNagri-Layout? I find it difficult to distinguish the characters from one another.
I can then compare the Bolnagri-Layout with the Inscript-Layout and a possible OptHIN-layout.

Will this do?

@hurrdudd The character [a] has a frequency of ~30%, which is quite a lot. Do you think this is correct? There are many double strokes for a. Are they usually written that way? The most common bigramms are: 821045 a 648070 aa 493560 e 392033 ha 320197 ee 317772 k 313123 ra 264597 ar 211511 ka 20564...

but got: $ from aksharamukha import transliterate from: can't read /var/mail/aksharamukha I am so sorry for not being verbose. The snippet I wrote was meant to be saved as a Python file and executed. If you want to batch convert documents then use the following code. #!/usr/bin/python3 # coding: ut...

@hurrdudd I do not yet know, how often the letters on the shifted level are used for devanagiri-input. If the usage is more than 5% of all keys, a thumb shift (either to the left or the right of space) might be a possible solution: https://opt-in-layout.org/download/file.php?id=4 Please note that t...