[Rule] Try on Colab before posting a technical issue
There is simple, reproducible single-cell Colab Usage here: https://hf.co/hexgrad/Kokoro-82M#usage
You should at least get it working on SOME machine (aka Google's machine), before you start complaining in this community tab that it does not work on YOUR machine.
If the copy-paste Usage cell does not work on Colab, it is DEFCON 1 and you should open a discussion immediately.
Otherwise, you need to show some effort to bridge the gap between Colab and your local setup, like at least put a couple reps in with a chatbot.
Discussions that fail to do this will receive a link to this pinned post and/or be immediately closed.
Does local deployment only support English?
Tried the collab code you linked to and getting these errors:
UnicodeDecodeError Traceback (most recent call last)
in <cell line: 0>()
3 # 🇫🇷 'f' => French fr-fr
4 # 🇮🇳 'h' => Hindi hi
----> 5 pipeline = KPipeline(lang_code='a') # make sure lang_code matches voice
6
7 # The following text is for demonstration purposes only, unseen during training
2 frames
/usr/lib/python3.11/encodings/ascii.py in decode(self, input, final)
24 class IncrementalDecoder(codecs.IncrementalDecoder):
25 def decode(self, input, final=False):
---> 26 return codecs.ascii_decode(input, self.errors)[0]
27
28 class StreamWriter(Codec,codecs.StreamWriter):
UnicodeDecodeError: 'ascii' codec can't decode byte 0xe2 in position 817: ordinal not in range(128)
@dstnrsh
I just ran it on Colab without issues. Without providing the text you used to trigger that error, it is unclear if I can reproduce, and therefore debug the issue. Unless you left the text exactly as-is, without editing after copy-paste?
Edit: It may have been a utf-8 encoding issue resolved by https://github.com/hexgrad/kokoro/pull/19#issuecomment-2622527804
Correct I just copied the code directly into a collab and ran it on a T4. I will try again.
Works great! Thank you!