RuntimeError: Could not infer dtype of numpy.float32

[YoungJay0612]

Hello, sorry to bother you. I have met a problem. I have configured the environment as described in the readme file, but the program is wrong when I run the run_predict_file.py, RuntimeError: Could not infer dtype of numpy.float32. Is the version of some libs are incorrectly? I have tried serveral versions of numpy, but it doesn't work and displays the same error.
Looking forward to your reply!

[gabrielmittag]

Hi, there is no run_predict_file.py, did you mean run_predict.py? Which arguments did you use? Here is an example on how to use it from the readme:
python run_predict.py --mode predict_file --pretrained_model weights/nisqa.tar --deg /path/to/wav/file.wav --output_dir /path/to/dir/with/results

[YoungJay0612]

Sorry，I have written wrong in the commnet. But actually I run the run_predict.py as the example told. Here is the results:

[gabrielmittag]

Hm I am not sure, could you share your wav file so I could test it? Or you could try one of the wav files from the NISQA dataset

[YoungJay0612]

test.zip
hello，thanks so much. the test.zip is the wav file I used, you can have a try. And I will ttry to test the files from the NISQA dataset to confirm.

[YoungJay0612]

Here is the result of the DEMAND data DLIVING_16k chn01.wav. The result is the same.

[gabrielmittag]

I just made a fresh install of the conda environment on an Ubuntu machine and cannot reproduce the error with the test.wav. Are you sure you are using the latest master branch? Did you install the environment with conda env create -f env.yml? Could you maybe double check which pytorch version you have installed in your environment?

[YoungJay0612]

As you mentioned, I have tried again. The version of the pytorch is as follows, and cannot find a version of 1.10.1 with pip install.

[gabrielmittag]

Hm that is kind of strange. I guess there must be something weird with your installation because the version should definitely be available with pip

[YoungJay0612]

It's really unimaginable. I reinstall the torch==1.10.1 today as what I have did yesterday, and sucess in one fell swoop. Now It can be running with your example script. Thanks a lot for your patient reply. Now I can test it any way I want.

[gabrielmittag]

No worries, great that it works for you now