Getting started

Installation

Build the maintained Arch package locally, or use the repository installer for a per-user source installation.

Arch package

AUR publication is pending

New AUR account registration is temporarily closed by Arch. The package recipe is complete and will be submitted when registration reopens.

Until then, build and install the same package definition directly from the repository:

git clone https://github.com/jweaver60/link-studio.git
cd link-studio
yay -S python-mediapipe
cd packaging/aur
makepkg -si

The package installs the application, desktop entry, AppStream metadata, bundled MediaPipe models, and setup helpers. python-mediapipe is sourced from the AUR; the remaining runtime dependencies come from the official Arch repositories.

Install from source

Clone the public repository and run the per-user installer:

git clone https://github.com/jweaver60/link-studio.git
cd link-studio
./scripts/install-user

The installer creates an isolated environment under ${XDG_DATA_HOME:-~/.local/share}/link-studio and installs launchers under ${XDG_BIN_HOME:-~/.local/bin}. It does not require root or modify desktop configuration.

Native requirements

For a source installation on Arch Linux, install the native GTK and GStreamer stack first:

sudo pacman -S gtk4 libadwaita python-gobject gstreamer \
  gst-plugins-base gst-plugins-good gst-plugins-bad gst-libav v4l-utils

The project launcher uses /usr/bin/python so the environment can access Arch's native GObject introspection bindings.

Optional components

Processed virtual camera

The virtual camera requires v4l2loopback-dkms, v4l2loopback-utils, and the headers for the running kernel. The helper installs the packages on Arch and loads a temporary device at /dev/video20:

link-studio-setup-virtual-camera

No persistent module configuration is written. After a reboot, run the helper again before enabling virtual-camera output.

Offline transcription

The local voice-note workflow can use whisper.cpp for transcription. This helper installs whisper-cpp when needed and downloads the multilingual base model:

link-studio-setup-local-ai

The model is approximately 142 MiB and is verified against a pinned SHA-256 digest before installation.

Verify the installation

link-studio --version
link-studio --diagnose

The diagnostic command is read-only. It reports detected devices, advertised formats, firmware information, safe control state, and the available audio/video stack.