# SuperCollider: scsynth, sclang and scide for sound synthesis and algorithmic composition > SuperCollider splits a real-time audio server from an interpreted language that controls it. That split is the whole design, and it decides who should install it. **supercollider/supercollider** — An audio server, programming language, and IDE for sound synthesis and algorithmic composition. - Repository: https://github.com/supercollider/supercollider - Website: http://supercollider.github.io - Stars: 6,751 · Forks: 854 - Language: C++ - License: GPL-3.0 - Published: 2026-09-22 · Updated: 2026-09-22 · Language: en - Canonical page: https://hysenlabs.com/projects/supercollider-supercollider ## What SuperCollider is, and who writes sclang SuperCollider is not one program. The README describes it as a platform made of four parts: scsynth, a real-time audio server with hundreds of unit generators for analysis, synthesis and processing; supernova, an alternative server with parallel DSP on multi-core processors; sclang, an interpreted language that controls the servers; and scide, an editing environment with an integrated help system. The audience named in the README is musicians, artists and researchers working with sound. That is a broad label, but the split tells you more. If you want to describe a sound as a graph of oscillators, filters and envelopes and then change that graph while it runs, the language is the interface. If you want a DAW timeline, this is the wrong shape entirely. The repository topics list livecoding alongside sonification, algorithmic-composition and computer-music, which matches the two groups who get the most from it: performers who modify code during a set, and researchers who need to generate sound from a script rather than perform it by hand. ## How scsynth, sclang and Quarks fit together The architecture is a client-server split with the language on the client side. sclang runs your code and sends instructions to scsynth or supernova, which owns the audio hardware and the DSP graph. That means the language can be restarted or edited without the audio thread being rewritten, and it means a crash in one half does not automatically take down the other. Unit generators are the building blocks inside the server; the README says there are hundreds of them, covering analysis, synthesis and processing. Third-party extension happens at two levels. On the server side, scsynth and supernova both expose C and C++ plugin APIs, so new UGens are compiled native code. On the language side, sclang has its own package manager called Quarks, which distributes sclang code rather than compiled DSP. Keeping those two extension paths separate is a deliberate constraint: a Quark can add classes, patterns and helper functions, but it cannot add a new UGen without a compiled plugin. The servers are the part that must not stall, and the language is the part you are expected to rewrite constantly. ## Installing SuperCollider and running a first sound The README points macOS and Windows users at the downloads page for stable release builds, with separate README_MACOS.md and README_WINDOWS.md files covering usage and self-building. Linux is the awkward case: to get the latest stable version, Linux users need to build SuperCollider themselves or use the Flatpak build, per README_LINUX.md. Raspberry Pi and Bela have their own README files. The most recent release listed in the repository is Version-3.14.1 from 2025-11-24. Once installed, the entry point is scide, which opens a document you evaluate line by line. From there, the README points you at the built-in documentation browser in scide for the language reference, with the same docs online at doc.sccode.org, and lists external starting points: A Gentle Introduction to SuperCollider by Bruno Ruviaro, Eli Fieldsteel's video tutorials, the Getting Started tutorial, Nick Collins' tutorial, and SCCode.org for user-submitted examples. Those tutorials are where the first working sound comes from; the README itself gives no example expression to paste. ## Where SuperCollider gets in your way The install story is the first real limitation, and the README states it plainly rather than hiding it: Linux users who want the latest stable release must build it or take the Flatpak. Building means CMake, Qt and Boost, and the platform table shows how narrow the tested combinations are. Testing is done on Windows 10 64-bit with MSVC 2022, macOS 15 with Xcode 15.2, and Ubuntu 26.04 with gcc 16. Guaranteed support covers Windows 10 and 11, MSVC 2022 and 2026, macOS 15 and 26, Xcode 16 and 26, Debian 11 and later, Ubuntu 24.04 and 26.04, Fedora 40 to 44, Arch Linux, gcc 13 and later, clang 18 and later, and Qt 6.2 and later. Qt 5.15 has only limited support. If your distribution or compiler falls outside that list, you are on your own in a C++17 codebase. The second limitation is the extension model. A musician who wants to ship a plugin for other people's DAWs will not find that here; the C and C++ APIs are for UGens loaded into scsynth or supernova, not a cross-host plugin format. The third is the interaction model itself. Everything happens by evaluating code, so there is no undo history to fall back on and no piano-roll editor to nudge a note two milliseconds later. ## SuperCollider against Pure Data and Max The closest alternatives are visual dataflow environments for the same job. Pure Data and Max both let you patch signal graphs by dragging boxes and connecting them, and both keep the audio engine and the patching layer separate in the same spirit as scsynth and sclang. The difference is the interface and what it optimizes for. In a patcher, the graph is the program and you read it spatially; in SuperCollider, the graph is written as text and you read it as code. That matters at scale. A hundred-node patch becomes a canvas you scroll around, while a hundred-line sclang routine can be generated by a loop, stored in version control, and diffed like any other source file. It also matters for live performance: typing a new value into a running expression is a different gesture from moving a slider, and the livecoding topic in the repository points at the crowd that prefers the former. Pure Data has the advantage of a much smaller footprint and no language runtime to learn, which is why it shows up in embedded and teaching contexts. If you already think in patches, SuperCollider will feel like relearning the same ideas through a narrower door. ## Maintenance, licence and the cost of upgrading The repository is not archived and the last push was on 2026-09-21, so work is ongoing. The release cadence visible in the repository is slow and deliberate: Version-3.14.1 arrived on 2025-11-24, preceded by two release candidates on 2025-11-16. That pattern, rc1 and rc2 before a final tag, tells you upgrades are staged rather than continuous, which is good news if you build from source and bad news if you were hoping for weekly fixes. The licence is GPL-3.0, and the README states SuperCollider is free software under Version 3 of the GNU General Public License with details in COPYING. For most users of the application this is unremarkable. For anyone who wants to link the server or the language into a closed-source product, GPL-3.0 is the constraint that decides the question, and the details belong with a lawyer rather than with this article. Upgrading from source also means tracking the platform support table: the guaranteed combinations shift between releases, and a compiler that worked for one version can fall outside the tested set for the next. ## Conclusion Adopt SuperCollider if you write code to generate sound and want a server you can drive from a live language session; the macOS and Windows downloads make that a short path. Do not adopt it if you need a drag-and-drop DAW workflow or a stable plugin format for a commercial release, since the C and C++ plugin APIs are not a substitute for VST or AU. Before committing, verify the platform support table against your exact OS and compiler, and check whether your target Linux distribution has a packaged build or only the Flatpak. ## FAQ ### What is SuperCollider used for? The README describes it as a platform for audio synthesis and algorithmic composition, used by musicians, artists and researchers working with sound. It covers audio analysis, synthesis and processing through the unit generators in scsynth and supernova. ### How do I install SuperCollider? macOS and Windows users get stable release builds from the downloads page, with usage and build instructions in README_MACOS.md and README_WINDOWS.md. Linux users need to build it themselves or use the Flatpak build, as described in README_LINUX.md. ### How do I use SuperCollider? You write code in sclang, the interpreted language, and it controls the scsynth or supernova audio server. The scide editing environment has a built-in documentation browser, and the same docs are online at doc.sccode.org. ### What is SuperCollider software? It is four components together: scsynth, a real-time audio server with hundreds of unit generators; supernova, an alternative server with parallel DSP on multi-core processors; sclang, the interpreted language that controls the servers; and scide, the editing environment. ### Is SuperCollider music or audio software? Both descriptions apply, since the README frames it as a platform for audio synthesis and algorithmic composition used by musicians, artists and researchers. The repository topics include computer-music, electronic-music, livecoding and sonification alongside audio. ## Sources - [License: GPL-3.0](https://github.com/supercollider/supercollider/blob/develop/LICENSE) - [Project website](http://supercollider.github.io) - [README](https://github.com/supercollider/supercollider/blob/develop/README.md) - [Releases](https://github.com/supercollider/supercollider/releases) - [supercollider/supercollider on GitHub](https://github.com/supercollider/supercollider) --- Hysen Labs editorial analysis, written from the project's own repository and release notes. Cite the canonical page: https://hysenlabs.com/projects/supercollider-supercollider