Closed happyprotean closed 5 years ago
Hi @happyprotean ,
CUR_X_FILES
is a variable containing names of files and CUR_X
the associated pre-processed mel-spectrogram. It seems the filename 'work\00097e21.wav' present in your stack-trace is not in CUR_X_FILES
.
Can you please provides first values of CUR_X_FILES
? I guess there is a mismatch in the file location stored in CUR_X_FILES
and the one requested by fastai library to open_fat2019_image
.
I tested exclusively under Linux, and you are using Windows, maybe there is some wizardry to support use of '\' instead of '/' in paths. :smile:
Thanks for your reply, sir. Yes, this error is caused by the difference between linux and window. Actually after I change from "fname = fn.split('/')[-1] " to "fname = fn.split('\')[-1] ", it could work. One more question, how long you spend on training this model?
You are welcomed. Congratulations to make it work under Windows. :+1:
Training took about one day with the hardware indicated in README.md.
Hi Sir, I met an error when I was trying to run training-CNN-model1 at step: Train on noisy data. It shows it cannot find files in the WORK folder. But I check the WORK folder, there is nothing. Do you know what may cause this error? Thanks a lot!
ValueError Traceback (most recent call last)