Arqenne Studio on Linux

Arqenne Studio runs on Linux with the same writing, narration and storyboard tools as on the Mac. Your projects live on your own computer, and the on-device AI works without an internet connection once its models are downloaded.

Choose your edition

Not sure? Pick Standard. It works everywhere, including PCs with AMD or Intel graphics. If you have an NVIDIA card, the NVIDIA edition makes narration with cloned and fine-tuned voices dramatically faster. Installing one edition replaces the other, and your projects and models are kept, so you can change your mind later.

Requirements

Both editions

• A 64-bit Intel or AMD processor (x86_64)
• Ubuntu 24.04 LTS or newer, or Debian 13. Ubuntu 22.04 and older are not supported.
• 8 GB of memory. Voice cloning needs at least 12 GB; 16 GB is recommended.
• About 20 GB of free disk space for the app and its on-device models

NVIDIA edition, additionally

• An NVIDIA GeForce RTX 20-series card or newer, or a comparable workstation card (Turing through Blackwell / RTX 50-series)
• 8 GB of graphics memory or more; voice cloning uses about 6 GB
• NVIDIA driver 580 or newer

Install on Ubuntu or Debian

From Arqenne's repository (recommended)

Add the repository once, and Arqenne installs and updates through apt like everything else on your system.

1. Add Arqenne's signing key:

curl -fsSL https://releases.arqenne.com/apt/arqenne-archive-keyring.gpg | sudo tee /usr/share/keyrings/arqenne-archive-keyring.gpg >/dev/null

2. Add the repository:

sudo tee /etc/apt/sources.list.d/arqenne.sources >/dev/null <<'EOF'
Types: deb
URIs: https://releases.arqenne.com/apt
Suites: stable
Components: main
Architectures: amd64
Signed-By: /usr/share/keyrings/arqenne-archive-keyring.gpg
EOF

3. Refresh the package list:

sudo apt update

4. Install Arqenne:

sudo apt install arqenne

For the NVIDIA edition, do two things first, then sudo apt install arqenne-cuda instead.

The driver and NVIDIA-repository commands below are written for Ubuntu 24.04. On Debian 13 or a newer Ubuntu release, install the driver as described on Debian's NvidiaGraphicsDrivers wiki page (or your Ubuntu release's driver tool), and pick the matching repository for your system on NVIDIA's CUDA downloads page. Everything else on this page applies to both.

1. Install the NVIDIA driver (skip this if nvidia-smi already shows driver 580 or newer):

sudo ubuntu-drivers install

sudo reboot

After the restart, run nvidia-smi. It should list your card, with a driver version of 580 or higher and a CUDA version of 13.0 or higher.

2. Add NVIDIA's software repository (once per computer). The NVIDIA edition uses NVIDIA's CUDA libraries, which Ubuntu and Debian don't include:

wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2404/x86_64/cuda-keyring_1.1-1_all.deb

sudo dpkg -i cuda-keyring_1.1-1_all.deb

sudo apt update

sudo apt install arqenne-cuda then also downloads the NVIDIA libraries it needs, about 3 GB.

From the download page

If you'd rather not add a repository:

• Standard: Download Arqenne for Linux (.deb)
• NVIDIA: Download Arqenne for Linux — NVIDIA (.deb) — install the NVIDIA driver and add NVIDIA's repository first, as above.

Then, in the folder you downloaded it to, install it by its file name. A file saved from the links above is usually named Arqenne.deb or Arqenne-cuda.deb; a file from a versioned link is named like Arqenne_1.3.1_amd64.deb. Use the name of the file you downloaded:

Standard:

sudo apt install ./Arqenne.deb

NVIDIA:

sudo apt install ./Arqenne-cuda.deb

apt installs anything else Arqenne needs. A copy installed this way updates itself from inside the app (see Updating).

Switching editions: installing one edition replaces the other, whether through the repository or from a download. Your projects and downloaded models are kept.

Arch Linux

An Arch package (AUR) is in preparation. Until it lands, Arqenne installs on Ubuntu and Debian only.

Start Arqenne

Open Arqenne from your applications menu, or run arqenne in a terminal. It's the same command on both editions.

First run

1. Sign in with your Arqenne account, or create one.
2. Download the on-device models from Settings › Models, or from the prompt on the home screen. All of them together are about 13 GB:
   • Local AI Assistant: chat and writing help that runs on your computer
   • AI Embedding: lets the assistant search your project's knowledge base
   • Standard Text-to-Speech: 28 built-in narration voices
   • Enhanced Text-to-Speech: voice cloning and fine-tuned voices (Creator and Studio plans; needs 12 GB of memory)

You only download each model once.

Check that your graphics card is being used (NVIDIA edition)

Generate a line of narration with a cloned or fine-tuned voice, then run:

nvidia-smi

Arqenne should appear in the process list, using several gigabytes of graphics memory.

Updating

Arqenne tells you when a new version is available. How it gets installed depends on how you installed it — Arqenne never replaces files behind your package manager's back.

• Installed from Arqenne's repository: update with the rest of your system, sudo apt upgrade, or from Software Updater. The app will name that command in its notice.
• Installed from a downloaded .deb: the notice has an Install and restart button. It asks for your password once, installs the new version and relaunches. You can also download the new package and install it the same way as before.

Your projects, settings and models are kept across updates.

Going back a version (repository installs): every published version stays available.

To stay on 1.3.0, for example:

sudo apt install arqenne=1.3.0 && sudo apt-mark hold arqenne

For the NVIDIA edition:

sudo apt install arqenne-cuda=1.3.0 && sudo apt-mark hold arqenne-cuda

To resume updates, unhold the package you held:

sudo apt-mark unhold arqenne arqenne-cuda

Verify your download

Every package is signed. Installs through apt are verified automatically; for a downloaded .deb, a SHA256SUMS file (and a signed SHA256SUMS.asc) is published next to each version at

https://releases.arqenne.com/download/<version>/linux-x86_64/SHA256SUMS

where <version> is the release number, for example 1.3.1. Save it in the same folder as your download, then run:

sha256sum -c --ignore-missing SHA256SUMS

It should print OK for each file it finds. The list names the versioned files (Arqenne_1.3.1_amd64.deb, Arqenne-cuda_1.3.1_amd64.deb), so rename a download saved as Arqenne.deb or Arqenne-cuda.deb to its versioned name before checking. The release key is Arqenne Releases <releases@arqenne.com>, fingerprint C6DA C5D8 6EAF 9E0E 2DC9 39AD 747A BE81 4ABB 61F3.

Uninstall

Standard edition:

sudo apt remove arqenne

NVIDIA edition:

sudo apt remove arqenne-cuda

Removing the app keeps Arqenne's repository and signing key. If you added the repository and want it gone too, delete its source file and let apt autoremove take the signing key (it was installed as the arqenne-archive-keyring package):

sudo rm /etc/apt/sources.list.d/arqenne.sources

sudo apt autoremove

Uninstalling keeps your work. To remove everything:

• Your projects are in ~/arqenne/studio/
• Downloaded models are in ~/.local/share/com.arqenne.studio/

Delete those folders only if you no longer need them.

Troubleshooting

Voice cloning doesn't appear Voice cloning needs a Creator or Studio plan and a computer with at least 12 GB of memory. It also needs the Enhanced Text-to-Speech model from Settings › Models.

The NVIDIA edition is as slow as the Standard edition

• Run nvidia-smi: the driver must be version 580 or newer.
• Check that NVIDIA's libraries installed: apt policy libcudnn9-cuda-13 should show an installed version. If not, repeat step 2 of the NVIDIA install.
• Start Arqenne from the applications menu or with arqenne.

You have more than one NVIDIA card Arqenne uses the first card, as numbered by nvidia-smi. To choose another, start it with, for example:

CUDA_VISIBLE_DEVICES=1 arqenne

No sound when playing narration Arqenne plays audio through PipeWire, ALSA or PulseAudio. Standard Ubuntu desktops already have this. On a minimal system, install pipewire-audio or alsa-utils.

apt install arqenne-cuda reports missing cuda-* packages NVIDIA's repository isn't set up yet — do step 2 of the NVIDIA install, then try again.

Same app as on the Mac

Everything from writing and the AI assistant to narration, voice cloning, storyboards and video export works the same on Linux as on the Mac. Updates are the exception: they arrive through apt or the in-app button, as described in Updating. Storyboard images render in Arqenne Cloud, or on your own ComfyUI if you connect one in Settings › Models, on both platforms.