Following this guide will allow you to get the best possible performance on Linux for professional audio needs. Even though these steps are well-tested, it is wise to research what each step accomplishes and why. See also https://wiki.archlinux.org/title/Professional_audio.

Note for users of other distros: Much of this guide can be adapted for other distros by simply switching out the package manager commands. For manually adding realtime privileges in other distros see jackaudio.org. In terms of kernels, Arch-based distros add the full preempt patch as part of the kernel config at build time whereas for others you might need to add preempt=full as a kernel parameter as part of step 4. From the yabridge documentation:

If the uname -a output contains PREEMPT_DYNAMIC, then run either zgrep PREEMPT /proc/config.gz or grep PREEMPT "/boot/config-$(uname -r)" depending on your distro. If CONFIG_PREEMPT is not set, then either add the preempt=full kernel parameter or better yet, switch to a kernel that's optimized for low latencies.

However, for Debian, Ubuntu or Arch, the low-latency Liquorix kernel might be an even better option. Install via:

curl -s 'https://liquorix.net/install-liquorix.sh' | sudo bash

With all other tweaks set identically, the Liquorix kernel performs noticeably better than the Debian kernel on my system. With Liquorix, there is no need to set preempt=full but the threadirqs parameter is still of great benefit.

If you want to get up and running quickly, these are the highest-impact steps:

1. Step 3 — Install realtime-privileges and add your user to the realtime and audio groups
2. Step 4 — Add the threadirqs kernel parameter
3. Step 5 — Set CPU governor to "performance" and disable sleep/screen locking

If you need to run Windows VST plugins, also follow Step 11 (yabridge). If you are still experiencing audio performance issues after these steps, work through the full guide below.

Pipewire is now the recommended audio server for pro audio on Arch. It handles ALSA, PulseAudio, and JACK compatibility in a single stack with lower latency overhead than the older JACK2 + PulseAudio combination.

Here's how to install pipewire if you are still running pulseaudio and jack:

yay -Rdd pulseaudio pulseaudio-alsa pulseaudio-jack jack2
yay -S pipewire-alsa pipewire-pulse pipewire-jack

Reboot then check whether pipewire and associated plugins are operational via:

inxi -Aa

For IRQ-based scheduling benefits when using ALSA, be sure to use the "Pro Audio" profile for your interface via your sound management tool.

JACK vs. pipewire-jack: For most DAW workflows, the native pipewire stack with the "Pro Audio" ALSA profile is the right choice. Use pipewire-jack (which provides a JACK compatibility layer) if you have software that explicitly requires a JACK server. You rarely need to run a separate JACK daemon anymore.

Optional: Install yay:

sudo pacman -S --needed git base-devel
git clone https://aur.archlinux.org/yay.git
cd yay
makepkg -si

In other Arch-derived distros, yay might already be installed or available via sudo pacman -S yay.

rtcqs audits your system configuration and reports which real-time audio settings are correctly configured. Run it both before and after applying the tweaks in this guide to verify improvements — green items indicate a passing check, red items indicate something that needs attention.

git clone https://codeberg.org/rtcqs/rtcqs.git
cd rtcqs
./src/rtcqs/rtcqs.py

yay -S realtime-privileges

Add your user to the realtime group (required for real-time scheduling) and the audio group (required to satisfy rtcqs):

sudo usermod -a -G realtime,audio $USER

Log out and back in, or reboot, for the group changes to take effect.

Note for non-Arch distros: If realtime-privileges is not available in your package manager, you can manually configure real-time limits by editing /etc/security/limits.conf and adding:

@audio   -  rtprio     95
@audio   -  memlock    unlimited

See jackaudio.org for details.

The threadirqs parameter moves IRQ (interrupt request) handling off the main CPU thread and into dedicated kernel threads. This reduces latency jitter caused by hardware interrupts competing with your audio workload, and is one of the most effective single tweaks for pro audio performance.

For GRUB:

sudo nano /etc/default/grub

Change GRUB_CMDLINE_LINUX="" to GRUB_CMDLINE_LINUX="threadirqs", then regenerate the GRUB config. On vanilla Arch:

sudo grub-mkconfig -o /boot/grub/grub.cfg

Some distros provide an update-grub wrapper that does the same thing.

For systemd-boot:

sudo nano /boot/loader/entries/arch.conf

Add threadirqs to the end of the options line, save, and reboot.

For EndeavourOS and unified kernel image setups:

sudo nano /etc/kernel/cmdline

Add threadirqs to the end of the options line, then run:

sudo reinstall-kernels

Set the CPU governor to "performance" to prevent the CPU from downclocking mid-session, which can cause audio dropouts. You can do this via your desktop environment. On KDE Plasma:

Alternatively, set it via the terminal:

sudo cpupower frequency-set -g performance

Or add cpufreq.default_governor=performance as a kernel parameter (note that this can be overridden by power management tools like Powerdevil).

Also disable sleep and screen locking while recording or mixing to prevent interruptions.

Laptop users: Running the performance governor continuously will increase heat and reduce battery life. Consider switching governors only during active recording/mixing sessions, or using a tool like auto-cpufreq to automate this.

Reducing swappiness tells the kernel to avoid swapping RAM to disk, which can cause unpredictable latency spikes.

sudo nano /etc/sysctl.d/99-sysctl.conf

Add:

vm.swappiness=10

If you run rtcqs.py and it gives you a warning about Spectre/Meltdown mitigations, you could add mitigations=off to GRUB_CMDLINE_LINUX.

Warning: Disabling these mitigations reduces your system's security against certain CPU-level exploits. Only consider this on a dedicated studio machine not used for general web browsing or untrusted code. See https://wiki.linuxaudio.org/wiki/system_configuration#disabling_spectre_and_meltdown_mitigations for more context.

yay -S base-devel

Pipewire users: If you are using the "Pro Audio" ALSA profile in pipewire, IRQ thread priority is handled automatically and this step may not be necessary. If you are not using pipewire, or if rtcqs still reports IRQ priority issues, install udev-rtirq.

git clone https://github.com/jhernberg/udev-rtirq.git
cd udev-rtirq
sudo make install
reboot

For regular work in a DAW, it is recommended to set its audio system to ALSA rather than going through pipewire or a JACK layer. Direct ALSA access bypasses the additional mixing layer, giving the lowest possible latency. If you need to monitor other audio sources during a session, in REAPER you can change the ALSA input and output devices to "default" (you need to type this manually):

Set your desired audio device using your desktop environment.

DAW Install Examples:

REAPER: http://reaper.fm/download.php or

yay -S reaper

Consider setting the RT (real-time) priority value to 80 on the audio device page. RT priority determines how aggressively the OS schedules the audio thread relative to other processes — higher values mean higher priority. A value of 80 matches the sane default used by Ardour and Mixbus, and is a reasonable starting point.

Ardour: https://community.ardour.org/download or

yay -S ardour

Bitwig Studio: https://www.bitwig.com/download/ (flatpak not compatible with yabridge) or

yay -S bitwig-studio

Also be sure to check out Qtractor, Tracktion Waveform, Mixbus, LMMS, Rosegarden, Zrythm etc.: https://en.wikipedia.org/wiki/List_of_Linux_audio_software#Digital_audio_workstations_(DAWs)

• My JSFX plugin collection (https://forum.cockos.com/showthread.php?t=275301)
• airwindows-git (http://www.airwindows.com/)
• x42-plugins (http://x42-plugins.com/x42/)
• lsp-plugins (https://lsp-plug.in/)
• zam-plugins (http://www.zamaudio.com/?p=976)
• distrho-ports (https://distrho.sourceforge.io/ports.php)
• dpf-plugins (https://distrho.sourceforge.io/plugins.php)
• elephantdsp-roomreverb (https://www.elephantdsp.com/)
• dragonfly-reverb (https://michaelwillis.github.io/dragonfly-reverb/)
• aether.lv2 (https://dougal-s.github.io/Aether/)
• Bertom Denoiser (https://www.bertomaudio.com/denoiser.html) (not in the Arch repos or AUR)
• sfizz / sfizz-git (https://sfz.tools/sfizz/)
• Chowdhury DSP (https://chowdsp.com/products.html) (available via AUR: yay chow)
• Auburn Sounds (https://www.auburnsounds.com/)
• TAL Software (https://tal-software.com/)
• Pianoteq (https://www.modartt.com/pianoteq)
• AudioThing (https://www.audiothing.net/)

Gentle Warning: Recent WINE development has introduced compatibility issues affecting some Windows plugins on Linux — common symptoms include plugin crashes on load, GUI rendering failures, and MIDI timing problems. For the most current compatibility information, check the yabridge "tested with" page before installing a particular Wine version. A better long-term approach is to migrate entirely to Linux-native applications and plugins. The main gap currently is spectral editing — there are no Linux equivalents to iZotope RX or Acon Digital Acoustica.

Perhaps start with vanilla wine-staging and see how you fare. If your workflows rely heavily on sample-heavy VSTi like Kontakt, you may find better performance with wine-tkg (fsync enabled).

yay -S wine-staging

Or install a specific version known to be compatible:

yay -S downgrade
sudo DOWNGRADE_FROM_ALA=1 downgrade wine-staging

Then pin the version to prevent it being upgraded:

# Add to /etc/pacman.conf:
IgnorePkg = wine-staging

Follow the instructions to git clone and install: https://github.com/Frogging-Family/wine-tkg-git/tree/master/wine-tkg-git#quick-how-to-

If using wine-tkg, set the WINEFSYNC environment variable to 1 according to https://github.com/robbert-vdh/yabridge#environment-configuration (depends on your display manager and login shell).

yay -S yabridge yabridgectl

Or for the latest git version:

yay -S yabridge-git yabridgectl-git

Note: Depending on your distro, you might need to enable the multilib repo first:

sudo nano /etc/pacman.conf

Uncomment these lines:

[multilib]
Include = /etc/pacman.d/mirrorlist

Configure yabridge according to https://github.com/robbert-vdh/yabridge#readme, then install your Windows VST2, VST3, or CLAP plugins.

Once everything is set up, verify that volume levels are set correctly. Run:

alsamixer

Check that output is set to 100 (vertical bars) or a gain of 0 dB (shown in the top left of alsamixer). Use F6 to select the correct sound card. You can also use your desktop environment's volume controls, though displayed numbers may not match alsamixer directly.

TIP: If you are finding that the volume of your interface is being reset to 40% on startup (Behringer UMC Series seems susceptible), switch to the "pro audio" profile in pavucontrol.

An xrun (overrun or underrun) occurs when the audio buffer is not filled in time, resulting in a click, pop, or dropout. To check whether you are experiencing xruns:

• REAPER: View > performance meter
• Ardour/Mixbus: The xrun counter is displayed in the top right of the window
• pipewire: Run pw-top in a terminal during a session to monitor real-time DSP load and missed deadlines
• JACK (if used): jack_iodelay and qjackctl's message window will report xruns

If you are consistently getting xruns, try increasing your buffer size first (e.g., 128 → 256 → 512 samples), then work back through this guide to identify the bottleneck.

Music Player: strawberry (can produce bit-perfect playback)

Tagger: kid3, picard
DDP creation/verification/etc: ddptools
Audio Converter: fre:ac or soundconverter
CD Ripper: asunder or cdrdao
CD Burner: cdrdao, k3b or nerolinux4