This project aims to create a modular synthesizer plugin (VST3, CLAP). Like those encountered in projects like VCVRack or Bitwig's Grid.
The core idea is having a hexagonal tile map for laying out module instances and connect them at the edges to route audio signals and control signals to inputs of other modules.
A goal is to provide a simple wireless environment to build sound effects, synthesizers or whole generative music patches from predefined modules.
Hosting plugins (VST, LV2, ...) is out of the scope of this project. The goal is rather to have a good set of predefined modules.
Here is a screenshot of how it looks:
Since June 2022 the project is under heavy development again. A rewrite of HexoTK took longer than anticipated, but provides all required features now. All development currently takes place on the "hexosynth_2022" branch. As of 2022-06-30 a raw skeleton of the original functionality is done. There is still probably more than 3-4 months until we are on the same feature level as HexoSynth was in 2021.
The "master" branch is still the state from 2021.
If you want to stay up to date, follow my devlog:
- https://m8geil.de/posts/hexosynth-1/
- https://m8geil.de/posts/hexosynth-2/
- https://m8geil.de/posts/hexosynth-3/
- for an up to date list, look here: https://m8geil.de/tags/hexosynth/
Make sure to follow Weird Constructors Mastodon account or the releases of this project to be notified once I release a beta or stable release.
If you want to do chat, feel free to join the RustAudio Discord / Community here: https://rust.audio/
The new branch 'hexosynth_2022' moves all GUI logic to the scripting language WLambda, which is an easily embeddable scripting language for Rust. All higher level functionality will be realised in WLambda. As well as the test suite for the GUI.
The Rust code contains all the low level functionality, such as the DSP code, the hexagonal grid data structure and the implementation of all the GUI widgets.
The scripting language code will be linked into the HexoSynth application at compile time. The goal is to deploy a single binary. But you could load the WLambda code from a different place using an environment variable later.
- A useable GUI with highly features parameter knobs and a hexagonal module/node matrix that is easily changeable with the mouse.
- Serialization/Deserialization of patches (even though the UI for patch/preset management has not been implemented yet, except a "Save" button).
- Signal monitors for the selected node.
- Per node and output signal feedback LEDs.
- A simple Tracker to edit "TSeq" sequences.
- A primitive sample selection browser.
And following DSP nodes:
| Category | Name | Function |
|---|---|---|
| IO Util | Out | Audio output (to DAW or Jack) |
| Osc | Sampl | Sample player |
| Osc | Sin | Sine oscillator |
| Osc | BOsc | Basic bandlimited waveform oscillator (waveforms: Sin, Tri, Saw, Pulse/Square) |
| Osc | VOsc | Vector phase shaping oscillator |
| Osc | Noise | Noise oscillator |
| Signal | Amp | Amplifier/Attenuator |
| Signal | SFilter | Simple collection of filters, useable for synthesis |
| Signal | Delay | Single tap signal delay |
| Signal | PVerb | Reverb node, based on Dattorros plate reverb algorithm |
| Signal | AllP | All-Pass filter based on internal delay line feedback |
| Signal | Comb | Comb filter |
| N->M | Mix3 | 3 channel mixer |
| N->M | Mux9 | 9 channel to 1 output multiplexer/switch |
| Ctrl | SMap | Simple control signal mapper |
| Ctrl | Map | Control signal mapper |
| Ctrl | CQnt | Control signal pitch quantizer |
| Ctrl | Quant | Pitch signal quantizer |
| Mod | TSeq | Tracker/pattern sequencer |
| Mod | Ad | Attack-Decay envelope |
| Mod | TsLFO | Tri/Saw waveform low frequency oscillator (LFO) |
| Mod | RndWk | Random walker, a Sample & Hold noise generator |
| IO Util | FbWr / FbRd | Utility modules for feedback in patches |
I have a pretty detailed TODO list in my private notebook, but this is the rough road map:
- DONE: Make a UI that is more or less fluently usable and easily extendable with new modules.
- DONE: Take a bit of care that there is online help.
- DONE: Factor out the DSP code into it's own crate.
- Redo the UI with performance optimized and overhauled HexoTK.
- Rebuild the UI logic of HexoSynth from 2021 with WLambda.
- Add preset/patch management to the UI.
- Add lots (many more than above listed) of modules (Oscillators, Filters, Envelopes, LFOs, Quantizers, ...).
- Add a MIDI-Ctrl interface for receiving pitch control signals, gate and clock from the DAW
- Add parameter input node for receiving automation from the DAW
- Add audio inputs for receiving audio from the DAW
- Comment the code for easier maintenance.
You might need following dependencies (Ubuntu Linux):
sudo apt install libjack0 libjack-jackd2-dev qjackctl libx11-xcb-dev
sudo apt install libxcb-icccm4-dev libxcb-dri3-dev
You might need following dependencies (Ubuntu 20.04 Linux):
sudo apt install libgl1-mesa-dev libjack-jackd2-dev qjackctl libxcursor-dev
sudo apt install libx11-xcb-dev libxcb-icccm4-dev libxcb-dri2-0-dev libxcb-dri3-dev
These might work on Debian too:
sudo apt install libjack0 libjack-dev libx11-xcb-dev libxcb-icccm4-dev libxcb-dri2-dev
Enter the nih_plug subdirectory:
hexosynth/$ cd nih_plug
Compile:
hexosynth/vst2/$ cargo xtask bundle hexosynth_plug --release
Install:
heyosynth/nih_plug/$ cp -vfr target/bundled/hexosynth_plug.vst3 ~/.vst3/
heyosynth/nih_plug/$ cp -vfr target/bundled/hexosynth_plug.clap ~/.vst3/
CPAL is a generic audio device abstraction library. It should work on most systems.
Enter the cpal_standalone subdirectory:
hexosynth/$ cd cpal_standalone
Compile and run:
hexosynth/cpal_standlone/$ cargo run --release
JACK Audio Connection Kit is a sound server API, which allows multiple audio applications to communicate with each other.
Enter the jack_standalone subdirectory:
hexosynth/$ cd jack_standalone
Compile and run:
hexosynth/jack_standlone/$ cargo run --release
As of 2022-07-04 HexoSynth has been tested with:
- Ubuntu Linux 20.04 and Bitwig: Works
- The ones you encounter and create as issues on GitHub.
I currently have a quite precise vision of what I want to achieve and my goal is to make music with this project eventually.
The projects is still young, and I currently don't have that much time to devote for project coordination. So please don't be offended if your issue rots in the GitHub issue tracker, or your pull requests is left dangling around for ages.
I might merge pull requests if I find the time and think that the contributions are in line with my vision.
Please bear in mind, that I can only accept contributions under the License of this project (GPLv3 or later).
If you want to help this project:
- Samples: Find drum or synth samples that I can publish in HexoSynth. I would love to provide a drum kit with default sounds that is always available.
- Binaries: Build Windows and/or Mac OS binaries: I've decided that I won't build windows binaries anymore. I don't like supporting the whole Apple & Micosoft vendor lock-in directly. However, I would be fine putting links to your binaries on my release pages or the HexoSynth README. Given that the binaries are current enough.
- More DSP nodes/modules: Implement more DSP nodes: You could extend HexoDSP
with new DSP implementations. Just reach out to me via Discord/IRC/Mastodon or Github
issue to discus the name of the new node. I currently don't have a guide for this,
but be aware that new DSP nodes need to meet these quality guidelines:
- Clean Rust code that I can understand and maintain.
- Does not drag in huge dependency trees. One rationale here is, that I don't want the sound of a HexoSynth patch to change because some upstream crate decided to change their DSP code. To have optimal control over this, I would love to have all the DSP code contained in HexoDSP. Make sure to link the repository the code comes from though.
- Come with automated smoke tests like all the other nodes, most test signal min/max/rms over time, as well as the frequency spectrum where applicable.
- It's parameters have proper denormalized mappings
(like
0.5 => 4000 Hzor0.3 => 200ms). - Provide short descriptions for the node and it's parameters.
- Provide a comprehensive longer help text with:
- What this node is about
- How to use it
- How the parameters work in combination
- Suggestions which combinations with other nodes might be interesting
- If applicable: provide a graph function for visualizing what it does.
You can reach me via Discord ( WeirdConstructor#7936 ), Mastodon (
@weirdconstructor@mastodon.online ) or IRC. I'm joined most public Rust Discord
servers, especially the "Rust Audio" Discord server. And I am also on IRC on
the network Libera.Chat in the #lad channel (nick wct).
If you don't have means to access any of them, you can alternatively send me a Github issue.
Don't use E-Mail, I only read them irregularly, and I might miss yours completely.
You can support me (and the development of this project) via Liberapay:
This project is licensed under the GNU General Public License Version 3 or later.
The fonts DejaVuSerif.ttf and DejaVuSansMono.ttf under the license:
Fonts are (c) Bitstream (see below). DejaVu changes are in public domain.
Glyphs imported from Arev fonts are (c) Tavmjong Bah (see below)
Picking a license for my code bothered me for a long time. I read many discussions about this topic. Read the license explanations. And discussed this matter with other developers.
First about why I write code for free at all, the reasons are:
- It's my passion to write computer programs. In my free time I can write the code I want, when I want and the way I want. I can freely allocate my time and freely choose the projects I want to work on.
- To help a friend or member of my family.
- To solve a problem I have.
- To learn something new.
Those are the reasons why I write code for free. Now the reasons why I publish the code, when I could as well keep it to myself:
- So that it may bring value to users and the free software community.
- Show my work as an artist.
- To get into contact with other developers.
- To exchange knowledge and help other developers.
- And it's a nice change to put some more polish on my private projects.
Most of those reasons don't yet justify GPL. The main point of the GPL, as far as I understand: The GPL makes sure the software stays free software until eternity. That the end user of the software always stays in control. That the users have the means to adapt the software to new platforms or use cases. Even if the original authors don't maintain the software anymore. It ultimately prevents "vendor lock in". I really dislike vendor lock in, especially as developer. Especially as developer I want and need to stay in control of the computers and software I use.
Another point is, that my work (and the work of any other developer) has a value. If I give away my work without any strings attached, I effectively work for free. This compromises the price I (and potentially other developers) can demand for the skill, workforce and time.
This makes two reasons for me to choose the GPL:
- I do not want to support vendor lock in scenarios for free. I want to prevent those when I have a choice, when I invest my private time to bring value to the end users.
- I don't want to low ball my own (and other developer's) wage and prices by giving away the work I spent my scarce private time on with no strings attached. I do not want companies to be able to use it in closed source projects to drive a vendor lock in scenario.
We can discuss relicensing of my code or project if you are interested in using it in a closed source project. Bear in mind, that I can only relicense the parts of the project I wrote. If the project contains GPL code from other projects and authors, I can't relicense it.