Tonal languages have major mispronunciations

[joseph2mi]

Path: /speech-synthesis/prompting

Unlike English, you cannot prompt "correct" pronunciations on languages like Chinese or Vietnamese. No matter how close the voices may sound, Microsoft Azure voices will still beat Eleven labs for the sole fact that the speech generated often messes up tonal markers when Azure doesn't. This is an important training base that needs to be fine-tuned.

[louisjoecodes]

Hi @joseph2mi have you tried our pronunciations feature? Could you give us some examples? Thanks!