Linux Audio Plugin Development (LAPD)
x42 (Robin Gareus)
Linux Audio developer interview with x42 (Robin Gareus)
This interview was conducted by Amadeus Paulussen in 2025.
Dear Robin, thank you for taking the time to answer my questions. 🙂
Although most of us are probably quite familiar with your name, I would appreciate it if you could briefly summarize who you are and what you do in relation to Linux Audio.
While I am a GNU/Linux enthusiast since three decades, my Linux Audio journey only started in the mid 2000s, when I was challenged to produce a short-film soundtrack.
I quickly realized that no tools for that task existed on that platform. Others had already laid groundwork, so I began contributing to a variety of projects, while building tools needed to score soundtracks, and later, to create multi-media art. There is little Linux Audio infrastructure that I didn't have my fingers in at some point.
When the financial crisis in 2009 dried up funding for media art, I went back to school, and during the following years, co-organized five Linux Audio Conferences and, in 2012, the Computer Art Congress. Around that time I realized that while I was an accomplished musician, my skills are better suited to assisting others creating art. So eventually, I pivoted from making one-off tools to focusing on applications for a wider audience, earning a PhD in the process.
Can you also please tell us a little bit about what you've achieved so far (and what you're working on currently) in relation to your work on Ardour?
While I contributed to an award winning art piece over the years, Ardour has been a full time job for me since over a decade now. That involves everything from developing the project, to running the website and forum, and providing user support, alongside with Paul Davis. Getting a niche free/libre software project to sustain itself is quite the achievement in itself.
In relation to that, I do regular contract work on derivative projects for a large US based console manufacturer, which also contributes to Ardour.
I still enjoy DSP for the fact that, as opposed to a moving target DAW, creating plugins is a project that can be completed. Lately, I work on the side with a Berlin based mastering engineer on a set of plugins.
I wonder, what are your thoughts on Ardour's distribution model? Personally, I didn't realize for quite some time that, in addition to the "unofficial" way of installing Ardour directly via Pacman (on Arch Linux or as a Flatpak, there are also official, paid, so-called “Ready-To-Run” binaries available, and that these would actually be the proper way of installing Ardour. 🫣
Musicians or engineers just want the convenience to get a ready-to-run version rather than building it themselves. Due to the complexity involved when building Ardour, third party builds are often substandard, and hence we only provide support for the official binary.
Compiling Ardour requires significant expertise, and the build-service is one way to raise revenue for the project.
I think it's a great model, which I also use for the x42-plugins. You can get the source-code, study it, build it yourself, and also help others; but convenience, quality assurance, and support comes at a financial cost.
Personally, I would love to live in a world where everything is Open Source, not only software. However, the reality is, I also use a lot of proprietary tools. Anyway, I often ask plugin vendors whether they could imagine making their products Open Source. Unfortunately, the vast majority categorically reject this idea. What is your view on this? Do you think Open Source is a viable business model for a plugin vendor? And what do you think of proprietary software in general?
I grew up with audio hardware that came with schematics, and believe that empowering the interested by providing access to the source is the right way forward. The valuable part is not intellectual property.
But most musicians don't know or care, and neither do crackers: one can find x42-plugins on warez sites with a keygen (!), while one could just compile them from source instead.
I make an income by providing support to users who care, and not by pursuing people willing to steal.
Do you prefer one plugin standard over another, or, to put it bluntly, could you rank LV2, CLAP, and VST3 in order of your preference?
Overall, LV2 sucks least; it features clean DSP/UI separation, message-passing, expressive semantic meta-data, and follows a DRY mindset.
I do like VST3, automation, and the fact that MIDI CCs are mapped as control parameters. This way, there is no ambiguity when a given parameter is automated and the host is in control.
CLAP took inspiration from LV2 and VST. It is a nice, extensible standard, and its killer feature is per note polyphonic automation and modulation.
The best designed plugin standard is Apple's Audio Unit, particularly for their variable I/O setup and callback architecture, but alas, it is non-free.
What is the reason, if I might ask, behind you only offering your x42-plugins as LV2?
Consistent experience is key. VST plugins would not have inline displays, for example.
There is also the historical aspect that I initially wrote those plugins only for myself, and LV2 is all that I need, and they are supported in all free/libre DAWs (and some non-free ones, too).
Do you have a favorite among the audio plugin frameworks, or do you perhaps even favor developing plugins without using a framework in the first place?
These days, I recommend FAUST and DPF.
Personally, I am very happy with my Linux-based computer life these days. And if there are things that are not quite ideal yet, it is mainly on the third-party software side of things. But, I must admit that, ever since I switched to Linux, I probably have had to learn more about computers than in my 25 years with Apple. 😜 However, as a result, I have never felt more connected to the software I use on a daily basis, primarily thanks to Open Source, actually speaking to developers, reporting issues, etc. How do you see the state of Linux Desktop and Linux Audio today? Are you generally satisfied, or are there still things that you think need optimization?
Linux Audio in general is doing amazingly well. There is hardly any audio hardware, embedded audio system, or an entertainment system that is not running Linux under the hood.
On the desktop, however, fragmentation is increasingly becoming an issue.
OK, a spicy one: Wayland, PipeWire, and Flatpak – blessings or curses? 🤓
Yes. :)
I have high hopes for PipeWire. It had a great start, yet repeated mistakes of earlier such projects by exposing yet another API (in addition to JACK, Pulse, and ALSA. Recent versions regressed to some extent, but the team is amazingly fast to respond and fix bugs.
Flatpak, however, is entirely unsuitable for pro-audio applications, not just because of the sandboxing, but also since it is conceptually at odds with 3rd party plugins.
It's also a step backward from having a shared userland, where one has to fix security issues only once for all applications. Yet, I acknowledge that it can be useful in some cases, notably for compatibility, and is closer to what people coming from an app-store platform are used to. But I wish others would stop shoehorning DAWs into Flatpak.
Wayland pretty much breaks everything, and I could rant for hours. It really solves no issues for me, yet breaks almost everything else. On top of it, performs worse than X, has no binary compatibility, and it is also a major problem for plugins since one cannot embed windows. I hoped it would fade and be abandoned after 10 years; maybe that still happens. :)
When I ran a web development studio in Switzerland, the developers who used Linux were known for constantly trying out new things, and occasionally breaking stuff as well. And while I myself like to be cutting edge and try out all kinds of things as well, I also value a simple, stable system. That's part of the reason why I decided from the outset to only use native plugins on Linux, and to see the limitations in that regard as an opportunity. What is your personal opinion on the use of Yabridge in connection with Windows plugin binaries on Linux?
I have no first hand user experience with Yabridge, and don't have any closed source binaries on my main system. But I hear that some Windows plugins run even better with Yabridge on Linux than they do on Windows, and I'm grateful to Robbert for having created the project.
Apart from software and, in our case, plugins, what do you think we could do to improve hardware support for audio interfaces and specialized devices and apps, such as speaker measurement, etc., for Linux?
Ideally vendors use standardized interfaces, and/or publish specs of their interface. USB-2 has been a great success there; thanks to Apple not allowing drivers on iOS, vendors just produced class-compliant devices, which then also work out-of-the-box on Linux. PreSonus publishes their control surface interface spec in the user manual. It would be great if other vendors would realize that their value is the actual hardware and that interface specs (and drivers) can be published.
Getting kernel driver and application developers NFR devices would be helpful as well.
Short of appealing to vendors, what we can do is chip in to buy certain devices and hand them to persons who will be grateful and write drivers in return.
As said, my Linux-journey so far has had a lot to do with learning, and for me personally, that wasn't just satisfying; it also felt natural. I mean, if I am eager to learn an instrument, for example, I have to put in the work as well. On macOS, on the other hand (I can't say much about the experience on Windows due to lack of experience), I at times got the impression that the learning curve was less steep, with the effect that not only myself, but probably most users, actually understand their systems only on a surface level. Yet, such users might argue that they can invest the time they save that way in creativity instead. How do you see this?
Don't let whatever you're trying to accomplish get in the way of deepening your knowledge about computer science.
After so many years in Open Source and Linux Audio, are you still motivated and positive about the movement, or has your enthusiasm waned somewhat over the years?
I would not call it a movement. Things have significantly changed in the last decade, and there are rock solid professional Linux Audio systems out there.
To answer your question, my enthusiasm changed. It didn't wane, but it matured alongside Linux Audio. :)
Ultimately, what would you suggest to vendors who already have plugins for Windows/macOS on the market and are considering Linux support?
It's fun to work on GNU/Linux.
On the practical side, develop on and target older LTS Linux distributions. Many users prefer stable distros for production, and cutting edge distros are backwards compatible.
If you are new to GNU/Linux, start with a dedicated Audio Distro (Ubuntu Studio, AVLinux, KXStudio. I also highly recommend reaching out to the community. Many devs are happy to share insights, notably to avoid pitfalls when building or CI setups, or distribution.
Last but not least, ideally, treat Linux plugins with the same care as for other OS and focus less on DRM, but provide quality products. From experience, I can tell that Linux users don't mind commercial products if they don't impose unreasonable constraints.
Finally, I would like to ask you what your hopes and fears are for the future of Linux Audio, and perhaps Open Source as well.
Free/libre software is already rather successful and takes the opposite route to being cross-platform. Many authoring tools (Blender, OBS, Audacity started on GNU/Linux, and now provide macOS and Windows versions. On the embedded side, for reliable systems, Linux is already the de-facto standard.
What I do fear, however, is that divergence keeps increasing, and we're already past the peak of Linux Audio on the desktop.
The time consuming part when creating plugins is honing parameter ranges, creating presets, tweaking UI layout, and listening. With more proprietary vendors entering the market, free/libre software volunteers may lose interest to continue if they cannot contribute.
On the upside, the ecosystem as a whole has improved, and it's easier than ever to professionally produce music on a GNU/Linux desktop, and I hope the trend continues.