SotF: an automatic eq for your speakers or headsets

The software helps you to get better sound from your speakers or your headsets. It can run as a TUI app (in a terminal) or a classical UI. What can you do with it?

• play music :)
• add audio plugins to customise the experience, an EQ, an upmixer for spatial audio or a binaural rendered? a limiter or a multiband compressor? a denoiser?
• create an EQ for your headphone!
• create an EQ for anechoic speaker measurements that you find on AudioScienceReview or ErinsAudioCorner (among others)
• create an room optimiser for your room, from simple to use to you are in control of every steps: a state of the art optimiser.
• compare one EQ with another. Customise to taste. Which one do you prefer?

Sound of the Future or SotF in short comes from the song from Giorgio Moroder made popular by Daft Punk. You can find many versions on Youtube. Here is an official one.

A picture is worth a thousand words.

Native Player

Terminal Player

How to use?

• Windows: download from Microsoft Store.
• Apple: download from the App Store.

Stable desktop releases are on the platform stores. Beta releases and command-line artifacts are on GitHub Releases, with another copy on our website.

If you like it, star the directory please. If you dont, please let us know why? All feedback is welcome: you can leave a comment on github or on AudioScienceReview.

Main functions

• An audio player (nothing special but useful to have bundle with the other applications)
• A audio recorder to measure your speakers
• A system to optimise the sound based on your measurements for speakers and from internet measurements for headset. We claim to have a SOTA optimiser (a bit early for it but improving month after month if you want to a engineer view of it).
• A large set of audio plugins that do spatial audio, binaural, denoiser, limiter, compressor, upmixer, equalizer, room optimiser, etc.
• A systemwide DSP that allow you to run the roomEQ system continously for all applications on your computer. Currently working well on MacOS and needs to be ported/tested on Linux and Windows.

The main UI application shows you the audio player. You can access the other functions via the menu bars.

For developpers

This is mainly a Rust application with some python and shell scripts. We want to keep it portable so we minimise C/C++ dependencies.

Cargo

Install rustup first.

If you already have cargo / rustup, you can jump to:

cargo install just
just

On Linux, select the correct install just command for your platform:

just install-...

On MacOS, select the correct install just command for your platform:

just install-macos

On Windows,

.\scripts\build-windows.bat

Then run post-install:

just post-install

Then download various data files

just download-once

You can build or test with a simple:

just build
just test

For QA there is a

just qa

that takes some time to run.

In order to build the audio player, we have 3 versions: a cli (command line interface), a tui (terminal UI) and a desktop.

cargo build --bin player-cli --release -p app-cli
just tui
just gpui

Since it is a rust stack, the binaries are generated in target/release/ directory and you can execute them from there.

In order to build signed binaries (required on MacOS): If you are a developper, you can self sign them with

./scripts/build-systemwide.sh
./scripts/build-dmg-sotf.sh

In order to publish them, you need to have an Apple developper ID:

./scripts/build-systemwide.sh --sign --notorized
./scripts/build-dmg-sotf.sh --sign --notorized

Where is the code?

The code is in four parts:

SotF

This repository sotf which is mostly an audio backend and the UI and TUI. The backend audio has a few components:

• an audio engine : an audio engine (process streams or files and output pcm to your audio device)

• an audio player: a library doing track management and 3 players, a CLI for testing, a TUI (terminal) based one and a desktop one with a native UI.

• a set of audio plugins:

  • host: a mini DAW that can run plugins in a list (like a rack) or in a graph (like a DAW) visualisation: loudness, spectrum, lufs
  • classical: iir and fir EQ, compressor (and multi-band compressor), limiter, gain, matrix, resampler, multi-band expander, convolution, delay, crossover, loudness compensation
  • spatial: upmixer from 2.0 to 9.1.6, binaural, cross-talk cancelation, mono to stereo, dowmixer.
  • denoiser, declicker, polyphonic note detection
  • a/b testing

Math related audio toolkit

math-testfunctions

A set of functions for testing non linear optimisation algorithms used in the next crate.

math-optimisation

A implementation of differential evolution algorithm (forked from Scipy) with an interface to NLopt and MetaHeuristics two libraries that also provide various optimisation algorithms. DE support linear and non-linear constraints and implement other features like JADE or adaptative behaviour.

Status: good for speaker equalisation. Not tested enough for other use cases.

math-iir-fir

An IIR and FIR filters implementation in Rust. Does what you expect. Compatible with Equalizer APO. It can generate various output formats.

Status: stable and working well.

math-dsp

A DSP library implementing test signals and fft based analysis.

Status: new.

math-solvers

A set of classical solvers with preconditionners that use LAPACK, BLAS and rayon for parallelisation. Support sparse matrices. Also can work in WASM which is convenient for web demos.

Status: correct and relatively fast but not optimised to death. WASM needs rust nightly to run in parallel.

math-wave

A set of functions to compute known analytical solutions of the wave equation.

Status: correct.

math-xem-common

Implement BEM and FEM for the Helmotz and wave equations. Support multigrid for both systems. XEM holds the common code.

Status: unknown, results match analytical results on simple mesh. Needs more testing especially for the advance features.

math-convexhull3d

This crate computes a convex hull in 3d.

Status: good quality aka no known bug.

math-bem

This crate implements a BEM (Boundary Element Method) solver, see BEM README

Status: ok-ish

math-fem

This crate implements a FEM (Finite Element Method) solver, see FEM README

Status: ok-ish

Automatic EQ

autoeq

Main crate with CLI binaries and optimization logic: it allow to optimise the sound from:

• autoeq: optimise anechoic data for speakers and data from a headphone
• convert-recording: automatically migrate old recordings to latest format.
• roomeq: optimise a set of speakers in room. The input and ouput formats are documented.

autoeq-cea2034

CEA2034/Spinorama calculations (listening window, early reflections, predicted in-room response, speaker scores)

autoeq-roomsim

WASM-targeted room acoustic simulator using BEM (boundary element method)

autoeq-datagen

Generate room data from the FEM and BEM package to test roomEQ.

GPUI toolkit

gpui-ui-kit

A comprehensive UI component library with 40+ components including:

• Core: Button, Card, Dialog, Menu, Tabs, Toast
• Forms: Input, NumberInput, Checkbox, Toggle, Select, Slider, ColorPicker
• Data Display: Badge, Progress, Spinner, Avatar, Typography
• Audio Controls: Potentiometer, VerticalSlider, VolumeKnob

See the gpui-ui-kit README for usage examples.

gpui-d3rs

A port of D3.js concepts to Rust with idiomatic builder patterns:

• Scales: Linear, Log with automatic tick generation
• Shapes: Lines, Bars, Areas, Arcs, Pies, Scatter plots
• Colors: RGB/HSL, interpolation, categorical schemes
• Geographic: Mercator, Orthographic projections
• Spatial: QuadTree, Delaunay triangulation, Voronoi
• Animation: Transitions, easing functions, timers

See the gpui-d3rs README for the full feature list and examples.

gpui-px

High-level charting API inspired by Plotly Express:

• 6 chart types: Scatter, Line, Bar, Heatmap, Contour, Isoline
• Fluent builder API
• Color scales: Viridis, Plasma, Inferno, Magma, Heat, Coolwarm
• Logarithmic scale support

See the gpui-px README for quick start examples.

FAQ

Why did you not reuse more code? The goal was to learn Rust and to learn other things I always wondered about:

• How to write an audio player? I took inspiration from camilladsp and wrote my own. I could have use Camilla (and I did at the beginning)
• Why are plotting library never perfect? I can usually go 90% of the way with most libraries but then I get block and then it gets complicated to get exactly what you want.
• Can I do everything in Rust from backend to fronted? I am not a fan of Typescript and the context switching between Rust and Typescript is not ideal for me. Using GPUI allow me to stay in Rust and be concentrated.
• Did LLM model progress enough to help building a complex app? Answer is yes since Opus 4.5 and Gemini 3.0. It is of course not perfect.
• Can I reuse my old c++ code with audio plugin? Answer is also yes, I translated most of them in Rust now. I am still unclear if I will be able to build AU plugins with GPUI but it is working for CLAP and VST3. I also get the plugins to work as AUplugin with some hacking and a bit of Swift.

MacOS specific: daemon and driver-hal

On MacOS the sound managent system does not make it easy to route audio between applications.

We added a few components:

• driver-hal: a HAL (Hardware Abstraction Layer) to route audio between applications. It is similar to BlackHole (https://github.com/BlackHoleSound/BlackHole) but it allow to route the audio through a DAW with plugins.
• daemon: a Rust deamon software that read audio from the HAL driver, apply a chain of plugins (typically to correct the sound of your headphone or software) and then send it to a hardware audio interface.
• toolbar: a software that allow to configure the daemon and HAL driver. It sits in the macos toolbar.

Status: almost working.

More pictures

Plugins list

EQ

Upmixer