Spleeter: [Discussion] I made model, so I wanna use it

[kno3a87]

spleeter train -p configs/jazz_config.json -d ./train/
I made model for piano, woodbass, drums for jazz music.

Question:

1. I can see "INFO:spleeter:Model training done" , but I cant find model
2. How can I use this model? please tell me command like spleeter separate -i audio_example.mp3 -o audio_output -p spleeter:4stems

[romi1502]

1. The model is saved in a directory set by the model_dir option in your json config_file (see the wiki).

2. You should then be able to use your trained model using spleeter separate -i audio_example.mp3 -o audio_output -p configs/jazz_config.json

[kno3a87]

1. I missed it! Thanks
2. I got this error

spleeter separate -i Autumn.wav -o Autumn_output -p spleeter/configs/jazz_config.json
INFO:spleeter:Downloading model archive https://github.com/deezer/spleeter/releases/download/v1.4.0/jazz_model.tar.gz
Traceback (most recent call last):
  File "/home/ubuntu/anaconda3/envs/spleeter-cpu/bin/spleeter", line 8, in <module>
    sys.exit(entrypoint())
  File "/home/ubuntu/anaconda3/envs/spleeter-cpu/lib/python3.7/site-packages/spleeter/__main__.py", line 47, in entrypoint
    main(sys.argv)
  File "/home/ubuntu/anaconda3/envs/spleeter-cpu/lib/python3.7/site-packages/spleeter/__main__.py", line 41, in main
    entrypoint(arguments, params)
  File "/home/ubuntu/anaconda3/envs/spleeter-cpu/lib/python3.7/site-packages/spleeter/commands/separate.py", line 166, in entrypoint
    estimator = create_estimator(params, arguments.MWF)
  File "/home/ubuntu/anaconda3/envs/spleeter-cpu/lib/python3.7/site-packages/spleeter/utils/estimator.py", line 36, in create_estimator
    params['model_dir'] = model_provider.get(model_directory)
  File "/home/ubuntu/anaconda3/envs/spleeter-cpu/lib/python3.7/site-packages/spleeter/model/provider/__init__.py", line 65, in get
    model_directory)
  File "/home/ubuntu/anaconda3/envs/spleeter-cpu/lib/python3.7/site-packages/spleeter/model/provider/github.py", line 65, in download
    raise IOError(f'Resource {url} not found')
OSError: Resource https://github.com/deezer/spleeter/releases/download/v1.4.0/jazz_model.tar.gz not found

[Faylixe]

1. I missed it! Thanks
2. I got this error

spleeter separate -i Autumn.wav -o Autumn_output -p spleeter/configs/jazz_config.json
INFO:spleeter:Downloading model archive https://github.com/deezer/spleeter/releases/download/v1.4.0/jazz_model.tar.gz
Traceback (most recent call last):
  File "/home/ubuntu/anaconda3/envs/spleeter-cpu/bin/spleeter", line 8, in <module>
    sys.exit(entrypoint())
  File "/home/ubuntu/anaconda3/envs/spleeter-cpu/lib/python3.7/site-packages/spleeter/__main__.py", line 47, in entrypoint
    main(sys.argv)
  File "/home/ubuntu/anaconda3/envs/spleeter-cpu/lib/python3.7/site-packages/spleeter/__main__.py", line 41, in main
    entrypoint(arguments, params)
  File "/home/ubuntu/anaconda3/envs/spleeter-cpu/lib/python3.7/site-packages/spleeter/commands/separate.py", line 166, in entrypoint
    estimator = create_estimator(params, arguments.MWF)
  File "/home/ubuntu/anaconda3/envs/spleeter-cpu/lib/python3.7/site-packages/spleeter/utils/estimator.py", line 36, in create_estimator
    params['model_dir'] = model_provider.get(model_directory)
  File "/home/ubuntu/anaconda3/envs/spleeter-cpu/lib/python3.7/site-packages/spleeter/model/provider/__init__.py", line 65, in get
    model_directory)
  File "/home/ubuntu/anaconda3/envs/spleeter-cpu/lib/python3.7/site-packages/spleeter/model/provider/github.py", line 65, in download
    raise IOError(f'Resource {url} not found')
OSError: Resource https://github.com/deezer/spleeter/releases/download/v1.4.0/jazz_model.tar.gz not found

This is actually a bug, sorry about that. We will fix it in the next release that should come next week. Meanwhile as a quick fix you can add an empty file with .probe name into your model directory. It will prevent _Spleeter_ from trying to download specified model from _GitHub_.

[kno3a87]

Then I add .probe, I can separate! Thanks!!!

[Faylixe]

We just released a new version that will simply create the probe file after the train command is executed so it will work as expected.