Replies: 1 comment
-
This repo is not maintained anymore, but I just fixed this in our fork (available via |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
I noticed that espeak-ng (1.52-dev) launched from the console produces wrong output (phonemization and sound). Some letters are completely omitted. The problem is already known: https://sourceforge.net/p/espeak/bugs/125/. Changing the character encoding in the console to UTF-8 (
chcp 65001
) does not help. The same thing happens with espeak_wrapper.py. For example: for polish word - "źrebię" ("foal" - in english), after running in python:result is:
"rˈɛbʲi"
but it should be:"ʑrˈɛbjɛ"
(referring to https://odo.lv/Espeak?eslang=-vpl&speed=160&output=-w&debug=-x&text=źrebię&formmethod=get)
The letters "ź" and "ę" are completely omitted. I think this also applies to languages other than polish. I'm not a python programmer, but checking def _espeak_exe in espeak_wrapper.py it looks like the text that is phonemized is readed from the console. I think this is the wrong approach and on windows it should be read from a UTF-8 encoded file, example:
espeak-ng -vpl --ipa -f utf8_text.txt
I modified espeak_wrapper.py (def phonemize_espeak) for my needs from:
to:
and now phonemize works properly. As I said, I'm not a python programmer, everything I did is based on google. Someone should take a closer look at this. I'm on windows 10, python v3.11, coqui TTS-0.22.0.
Beta Was this translation helpful? Give feedback.
All reactions