Linguistic grouping, non latin characters

[whinette]

I am sorry for the late answer, I got overworked.
There would be 96 main characters, 48 hiragana and 48 katana, both of these are quite different even if it is the same sound (basically, they exprime word from different provenance or concept).

For the variants:
The small っ (sokuon) can never be at the beginning of a word, we can exclude it from grouping.
The yoon are composite sound and can be ignored (collated into the first sound), for example しゃ would be into し).
The dakuten are more a problem: these are different sounds and different unicode character. There are 30 voiced hiragana and 30 voiced katakana.

The total would be ... at least 156 chars.

I think my proposal would be easier on you and have also the advantage to provide more flexibility for other users.