Getting MMMAudio working with Windows/WSL

Install WSL - You may want to follow current LLM instructions, but this worked for me:

Inside the PowerShell:

wsl --install
dism.exe /online /enable-feature /featurename:VirtualMachinePlatform /all /norestart

you may need to do this separately: Go to the Windows Store and get Ubuntu 22.04.5 LTS

in the PowerShell (not wsl) run:

wsl --version

which should give you an output like this:

WSL version: 2.6.3.0
Kernel version: 6.6.87.2-1
WSLg version: 1.0.71
MSRDC version: 1.2.6353
Direct3D version: 1.611.1-81528511
DXCore version: 10.0.26100.1-240331-1435.ge-release
Windows version: 10.0.26200.7623

and then:

wsl --list --verbose

which should give you something like this:

  NAME            STATE           VERSION
* Ubuntu-22.04    Running         2

Open Visual Studio Code and Install and Enable the WSL extension (in Extensions)

Go to Command Palette - Connect to wsl

Install the Mojo and Python extensions for WSL

open a new terminal in VSCode

this should open a wsl terminal

cd ~/

now we are in the user directory

clone the MMMAudio program

git clone https://github.com/spluta/MMMAudio.git
cd MMMAudio

check out the sounddevice branch (wsl doesn't like pyaudio):

git checkout --track origin/sounddevice

if you type:

git branch

the sounddevice branch should be selected

cd ~/

create the alsa config file and open it in vscode for editing:

code ~/.asoundrc

enter this text into the .asoundrc:

pcm.!default {
    type pulse
    hint.description "PulseAudio Sound Server"
}

ctl.!default {
    type pulse
}

pcm.pulse {
    type pulse
}

ctl.pulse {
    type pulse
}

create and edit the pulseaudio config file:

code ~/.config/pulse/daemon.conf

add this to the file and save it:

default-fragments = 5
default-fragment-size-msec = 2
nice-level = -20
high-priority = yes

install python3.13 (you can use 3.13 and above):

sudo apt update
sudo apt install build-essential

sudo apt install software-properties-common
sudo add-apt-repository ppa:deadsnakes/ppa
sudo apt install python3.13
sudo apt install pulseaudio-utils

I had to run this for pip to be installed:

sudo apt update && sudo apt install python3.13-venv

Now make and activate the virtual environment:

python3.13 -m venv venv

source venv/bin/activate

install the dependencies:

pip install numpy scipy librosa pyautogui torch mido python-osc matplotlib PySide6 hid sounddevice mojo==0.25.6.1

In VSCode: View->Command Palette-> Select Python Interpreter select the python inside your virtual environment

Open one of the examples and run it!

Some notes: - Mouse does not work. Use mmm_audio.fake_mouse() to fake the mouse with a GUI. - There is currently no MIDI working on windows/wsl. rt-midi doesn't build. Maybe someone can figure this out? - Open Sound Control works, but you have to send OSC messages to the ip address of wsl, which can be accessed this way:

hostname -I

Now go back to the Getting Started Guide and look at 3. and 4.