[Bug] Unable to succesfully run pip install spleeter

[Emaan-Confinality]

• I didn't find a similar issue already open.
• I read the documentation (README AND Wiki)
• I have installed FFMpeg
• My problem is related to Spleeter only, not a derivative product (such as Webapplication, or GUI provided by others)

Description

I think the new update 4 days ago related to compatible versions has caused some issue because previously the same code was working fine and giving output however now it is showing me numpy version error.

Step to reproduce

1. Running on colab.
2. Installed using !pip install spleeter
3. Got error:
   "Collecting spleeter
   Using cached spleeter-2.1.0-py3-none-any.whl.metadata (10 kB)
   Collecting ffmpeg-python==0.2.0 (from spleeter)
   Using cached ffmpeg_python-0.2.0-py3-none-any.whl.metadata (1.7 kB)
   Collecting httpx<0.17.0,>=0.16.1 (from httpx[http2]<0.17.0,>=0.16.1->spleeter)
   Using cached httpx-0.16.1-py3-none-any.whl.metadata (38 kB)
   Collecting librosa==0.8.0 (from spleeter)
   Using cached librosa-0.8.0.tar.gz (183 kB)
   Preparing metadata (setup.py) ... done
   Collecting norbert==0.2.1 (from spleeter)
   Using cached norbert-0.2.1-py2.py3-none-any.whl.metadata (3.8 kB)
   Collecting numpy<1.19.0,>=1.16.0 (from spleeter)
   Using cached numpy-1.18.5.zip (5.4 MB)
   Installing build dependencies ... done
   Getting requirements to build wheel ... done
   error: subprocess-exited-with-error

× Preparing metadata (pyproject.toml) did not run successfully.
│ exit code: 1
╰─> See above for output.

note: This error originates from a subprocess, and is likely not a problem with pip.
Preparing metadata (pyproject.toml) ... error
error: metadata-generation-failed

× Encountered error while generating package metadata.
╰─> See above for output.

note: This is an issue with the package mentioned above, not pip.
hint: See above for details."

Output

previously this command would run perfectly fine and install spleeter.

Environment

OS	Windows 11
Installation type	pip
RAM available	12.7 GB
Hardware spec	GPU / CPU both

[JoseffRibble]

This is likely due to the fact that distutils has been depreciated since python3.10 and the collab is running python3.11.

[opencoca]

@Emaan-Confinality the follow is not a fix but a work around using pyenv to install and older copy of python on collab:

!git clone https://github.com/pyenv/pyenv.git ~/.pyenv
import os

# Set the root for pyenv
os.environ['PYENV_ROOT'] = os.path.expanduser("~/.pyenv")
# Prepend pyenv's bin folder to PATH
os.environ['PATH'] = os.environ['PYENV_ROOT'] + '/bin:' + os.environ['PATH']

# Confirm pyenv is callable
!pyenv --version
!apt-get install libffi-dev
!pyenv install 3.8.13
!pyenv global 3.8.13
!pyenv exec pip install spleeter
!pyenv exec spleeter

Once this is done you should be able to call spleeter as needed with !pyenv exec spleeter as it will run using python 3.8.13.

You can build with it as you want just your notebooks will always have to use !pyenv to call it as though its on the command line (you could also use a subprocess but magic is more fun)

[cacaedu]

Hello.

Is there any way to fix this? Is it not possible to update outdated libraries?

Using old versions of Python or something similar doesn't seem like a good solution to me.

Thanks

[cacaedu]

This is likely due to the fact that distutils has been depreciated since python3.10 and the collab is running python3.11.

Hello @JoseffRibble . It's not just with collab. It's also happening at the local installation.

[JoseffRibble]

Unfortunately, until the devs update the code, using an older python version is the only solution I have. You can use something like pyenv to create a virtual environment with an older python version. This will allow you to run the code without having to modify your system's python version

[opencoca]

Yup, The PyEnv fix works everywhere. I even use it in my more esoteric docker builds where I need choose to combine the use of different python versions.

[HaoleiH]

I found a way to run it on colab.
The main issue is that cloab now is using python3.11. It provide a method to allow you temporarily switch back to python3.10 though. you can try it in this official colab notebook.

Make sure the notebook is connected to google server.

tools - command palette - type "Use fallback runtime version"

This will reset the python version to be 3.10 and everything works now.

According to colab official repository, this fallback runtime feature "is available until early February"

[opencoca]

As it's it's early February I strongly recommend using my pyenv workaround.

#927 (comment)

Using pyenv this way will allow you to have multiple versions of python on the colab and won't complicate things using a fallback that's about to be retired any day now.

[Emaan-Confinality]

for some reason, it is working fine on google vertex AI's colab enterprise notebook. Python version is 3.10.12.

This is likely due to the fact that distutils has been depreciated since python3.10 and the collab is running python3.11.

Hello @JoseffRibble . It's not just with collab. It's also happening at the local installation.

[opencoca]

@Emaan-Confinality, well yes that would be the reason: Python 3.10 should work with with spleeter. As @HaoleiH Google's depreciating this on the non enterprise colab this month. They'll likely keep it around for londer on enterprise.

That being said and me sounding like a broken record... just use pyenv wherever you're working with this as it's got slightly esoteric requirements and that will allow you to sidestep them causing other complications.