uroman phonemization
#1
by Aregay01 - opened
The characters ቸ and ጨ, ተ and ጠ, as well as አ and ዐ are mapped to the same phoneme, which introduces noticeable errors in the synthesized speech. This issue arises from inefficient phonemization that fails to adequately represent Tigrinya phonology. Are there any suggestions for expanding the phoneme inventory or improving vocabulary coverage?