I was planning to develop a language model for Devanagari script, but we have a lot of alphabets and variations to even try. क, ख, अ, आ and their variations such as के, का, खे खा, etc.
Also, we have half letters as well. For example, क् in क्या.
How do you guys propose to build alphabet.txt in the cases like these ?
I realized that, के is made with two unicode characters combined “क” and " े". Would it be okay to have it (के) as a alphabet in alphabets.txt ?
Any kind of lead would be helpful for me in this. Are there any other scripts like Devanagiri ?
I was also interested in something similar. I am not into programming n all but willing to learn.
Do you mean language model for DeepSpeech?