phonetics training - problem with superscript letters

[Klemich]

Your Feature Request

hi,

I'm currently trying a training set for phonetics. I'm first focussing on the languages I need to OCR, not the whole International Phonetic Alphabet, but soon why not.

Problems I've noticed so far concern superscript numbers and letter such as ʰ or ʷ.
My training set is probably too small, but the algorithm doesn't seem to be able to distinguish h from ʰ, w from ʷ.

The i next to a superscript number seems to blend with it, and gives a 1 and a superscripted dash.

Sample I need to OCR looks like that tsɔŋ²¹tsi²¹pʰui⁵³

Can you help me ?

Thanks