model error about speech HOT 8 CLOSED

This means your PYTHONPATH is not set correctly. I would double check that when you source setup.sh the paths are being set appropriately. In particular:

echo $PYTHONPATH and make sure that the path to the repository is there.

One possible error is that the source setup.sh must be done from the top level directory (speech/).

from speech.

when I exec "echo $PYTHONPATH", It tells me “..../speech-master/libs/warp-ctc/pytorch_binding: ", is it correct? @awni

from speech.

No it's not correct, it should be something like:

<path_to>/speech-master/..:<path_to>/speech-master/libs/warp-ctc/pytorch_binding:

I'm not sure why the setup.sh script didn't work for you. I'll try to find a way to reproduce this, but in the meantime, you can set the path manually for now or debug the setup.sh script yourself and submit a patch if you figure out the issue. The script is not complicated, just two lines.

from speech.

It is OK. ".../speech-master/..:/home/pupil/win/speech-master/libs/warp-ctc/pytorch_binding:". But it still t tells me "ImportError: No module named speech". @awni

from speech.

Oops, I get the problem now, the module name is the repo name (bad design!). If you change the repo from speech-master to speech it should work.

I will file an issue to fix the package structure so this does not break.

from speech.

sorry, I cannot catch you. I do not find that any folder named speech. can you tell me more Specific measures.

from speech.

when I exec "sudo make", it tells me "ImportError: No module named torch", it works good yesterday,
what happend? @awni

from speech.

If you change the name of the repository from speech-master to speech it should solve the import speech error.

from speech.