Skip to content

Fix for linux installation - #1068

Closed
mithusingh32 wants to merge 1 commit into
Anjok07:masterfrom
mithusingh32:fix_for_linux
Closed

mithusingh32 wants to merge 1 commit into
Anjok07:masterfrom
mithusingh32:fix_for_linux

Conversation

@mithusingh32

Copy link
Copy Markdown
  • Updated numpy version to 1.24 or higher
  • Removed Dora references
  • Update README

- Updated numpy version to 1.24 or higher
- Removed Dora references
- Update README
@ChrisTG742

Copy link
Copy Markdown

@mithusingh32 did you test that updating numpy does not cause incompatibilites? There was/is some problems with librosa and numpy: #1016
Or did you update librosa aswell?

@mithusingh32

mithusingh32 commented Jan 2, 2024 •

Copy link
Copy Markdown
Author

I did not update librosa.

I did some initially testing w/ demucs v4 htddemucs_ft with all default parameters and it worked.
I will try perform more testing with other models + parameters.

GPU: RTX 3070ti 8GB

@shervinemami

Copy link
Copy Markdown

Hi @mithusingh32 , I tried your Linux fix and it worked for me on Kubuntu 24.04. I tested with htdemucs_ft model on GPU as well as MDXNet using kuielab model on GPU. I had to rollback to using the older Python 3.11 instead of Python 3.12 that my system comes with.

I recommend adding this info to the Linux installation README so that all users can use the same Python version no matter their system comes with:

# Install UVR5 into a Python 3.11 venv (not virtualenv):
sudo add-apt-repository ppa:deadsnakes/ppa   # Access many Python versions
sudo apt update         # Allow installing many Python versions
sudo apt install python3.11        # Install Python 3.11
sudo apt install python3.11-venv   # So you can create venv virtual environments
sudo apt install python3.11-dev    # So you can build your own python packages, instead of only downloading them off the web
sudo apt install python3.11-tk     # So UVR can display graphics
python3.11 -m venv py3.11_UVR5.venv    # Create a new venv using Python 3.11
source py3.11_UVR5.venv/bin/activate
pip3 install --upgrade pip wheel   # Install or update the pip & wheel versions that are inside the venv
SKLEARN_ALLOW_DEPRECATED_SKLEARN_PACKAGE_INSTALL=True pip3 install -r requirements.txt

# Run UVR5 in the Python virtual environment:
source py3.11_UVR5.venv/bin/activate
python3 UVR.py

@Anjok07 Anjok07 closed this Dec 3, 2024
@GHDaddy

GHDaddy commented Aug 25, 2025

Copy link
Copy Markdown

Hi @mithusingh32 , I tried your Linux fix and it worked for me on Kubuntu 24.04. I tested with htdemucs_ft model on GPU as well as MDXNet using kuielab model on GPU. I had to rollback to using the older Python 3.11 instead of Python 3.12 that my system comes with.

I recommend adding this info to the Linux installation README so that all users can use the same Python version no matter their system comes with:

# Install UVR5 into a Python 3.11 venv (not virtualenv):
sudo add-apt-repository ppa:deadsnakes/ppa   # Access many Python versions
sudo apt update         # Allow installing many Python versions
sudo apt install python3.11        # Install Python 3.11
sudo apt install python3.11-venv   # So you can create venv virtual environments
sudo apt install python3.11-dev    # So you can build your own python packages, instead of only downloading them off the web
sudo apt install python3.11-tk     # So UVR can display graphics
python3.11 -m venv py3.11_UVR5.venv    # Create a new venv using Python 3.11
source py3.11_UVR5.venv/bin/activate
pip3 install --upgrade pip wheel   # Install or update the pip & wheel versions that are inside the venv
SKLEARN_ALLOW_DEPRECATED_SKLEARN_PACKAGE_INSTALL=True pip3 install -r requirements.txt

# Run UVR5 in the Python virtual environment:
source py3.11_UVR5.venv/bin/activate
python3 UVR.py

Dude idk what changed between my install and this, but doing this in a fresh venv worked. You saved me hours of ripping my hair out

@shervinemami

Copy link
Copy Markdown

Dude idk what changed between my install and this, but doing this in a fresh venv worked. You saved me hours of ripping my hair out

I'm glad to hear!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

5 participants