Readlink & Optimizer_Type errors when preparing the audio and running training script

[qunliu]

Thanks for sharing your great work!

Following the steps in the ReadMe, I have these error messages when preparing the audio and running the training script:

How do I solve the issues?

I am running the code on a machine with macOS Big Sur.

Also do you have ONNX version of your models to share? The main reason I was running the training code is trying to convert your PyTorch models to ONNX models, it'd help us a lot if you already have them available!

Thanks!

[qiuqiangkong]

Hi, just pull the latest code. The ./scripts/4_train/musdb18/train.sh is replaced by ./scripts/4_train/musdb18/configs/accompaniment-vocals,resunet_subbandtime.yaml. You will find the 'optimizer_type' flag.
Sorry but there is no ONNX supported now.

[qunliu]

Thanks for your quick reply, however I am still getting the same errors following the steps. I noticed the only change was a version change in setup.py, did I miss anything else?

[qunliu]

Oh, replacing with this line: TRAIN_CONFIG_YAML="scripts/4_train/musdb18/configs/vocals-accompaniment,unet.yaml" to
TRAIN_CONFIG_YAML="scripts/4_train/musdb18/configs/accompaniment-vocals,resunet_subbandtime.yaml" in the train.sh file fixes the "Optimizer_Type" error.

Now I am getting this error message:
"AssertionError: The directory ./workspaces/bytesep/evaluation_audios/musdb18 is empty!"

I think is related to readlink issue above, I am still trying to figure it out...

[qunliu]

Okay, replacing: MUSDB18_DATASET_DIR=readlink -f $MUSDB18_DATASET_DIR with MUSDB18_DATASET_DIR=greadlink -f $MUSDB18_DATASET_DIR also fixes the readlink issue.

Note the 'readlink' is replaced with 'greadlink' above, and you may use "brew install coreutils" to get the correct toolset installed, see this link for details: https://stackoverflow.com/questions/1055671/how-can-i-get-the-behavior-of-gnus-readlink-f-on-a-mac