Quickstart

This page assumes you already created the conda environment (see Installation).

1) Start the GUI

python patchbay_gui.py

2) Load an audio file

Go to Input & Anchors and load a WAV/MP3/etc.
You should see the waveform and be able to Play / Stop / Rewind.

3) Add a description

Switch to Description & Run, enter a short description of what you want to extract (e.g. “speech”, “drums”, “birds”), then pick a model from the drop-down.

First run may download model weights from Hugging Face. If the download fails due to gated access, see SAM-Audio Access Tokens.

4) Run

Click Run.
Use Abort to cancel an in-flight run.

5) Inspect outputs

Switch to Output to listen to target and residual, apply AudioFX, and export.