Install without Docker

AudioSilo also ships as one self-contained program for Linux, macOS and Windows. It suits a desktop machine or a small home server without Docker. The web player is built in.

tip

For an always-on, multi-user or reverse-proxied setup, Docker is the recommended path.

Download a release

Download your system's file from GitHub releases:

• a .tar.gz for Linux or macOS, or a .zip for Windows (amd64 or arm64);
• or a .deb or .rpm package for Linux, which also sets up a system service (see Linux packages and systemd).

checksums.txt lets you verify the download. Unpack the archive and you have one program, audiosilo.

Run it from a terminal

./audiosilo --data ./data

1. The first start prints the admin password and an auth code once. Save them. If you lose them, ./audiosilo --data ./data reset-admin sets a new password (see If you lose the password).
2. Open https://localhost:8080/admin. Expect a certificate warning at first; accept it.
3. Sign in as admin. The console walks you through setting up your server.

To choose the admin password in your browser instead, start with --setup. The server then prints a one-time link to a page where you create the admin account:

./audiosilo --data ./data --setup

Useful options:

Option	What it does
--data <dir>	Where settings, the database and certificates live (default ./data).
--setup	On first run, choose the admin password in the browser.
--ffmpeg <path>	Path to ffmpeg; "" turns off converting audio.
--ffprobe <path>	Path to ffprobe; "" turns it off. AudioSilo still reads most files' lengths and chapters itself.

Linux packages and systemd

The .deb and .rpm packages install audiosilo, pull in your system's ffmpeg, and add a systemd service. Create the service's user and data folder once, then start it:

sudo useradd --system --home /var/lib/audiosilo --shell /usr/sbin/nologin audiosilo
sudo install -d -o audiosilo -g audiosilo /var/lib/audiosilo
sudo systemctl enable --now audiosilo

The first-run password appears in the service's log:

journalctl -u audiosilo

Update

1. Download the new release from GitHub releases.
2. Stop the server.
3. Replace the old audiosilo program with the new one. With a package, install the new .deb or .rpm with your package manager instead.
4. Start the server again, with the same --data folder. For the service, run sudo systemctl restart audiosilo.

Your data folder and your books are not changed. When an update changes the database, the server first saves a backup of it. Server > About in the admin console tells you when a new version is out.

ffmpeg and ffprobe

AudioSilo uses ffprobe to read lengths and chapters, and ffmpeg to convert audio a browser can't play. They are not in the download, but you usually have nothing to do:

1. The server looks for them next to audiosilo, then on your PATH.
2. If it finds neither, it downloads a trusted copy once into tools/ in your data folder.
3. If that can't happen (offline, say), the server still runs. It doesn't convert audio, and the few files it can't read itself have no length or chapters. It tries again at the next start.

FFmpeg 7.1 or 8.1 makes first scans much faster. With those versions, or with no ffprobe at all, the server reads most files itself. With any other version, ffprobe reads every file. The copy the server downloads for Linux and Windows is 8.1. Run ffprobe -version to see which you have.

Good to know

Building from source

With Go 1.26 or newer and Node 24:

git clone https://github.com/KodeStar/audiosilo-server.git
cd audiosilo-server
scripts/build-admin.sh        # builds the admin console so it's included
go build -o audiosilo ./cmd/audiosilo

A source build does not include the web player, so /web stays off. The release downloads and the Docker image include it. See the developer docs for bundling it.

Licence

AudioSilo is licensed under AGPL-3.0. The Linux downloads also contain the GNU C Library (LGPL-2.1-or-later). Each archive includes LICENSE and NOTICE; the packages install NOTICE under /usr/share/doc/audiosilo/.