Speaker Sim turns a pair of measured loudspeakers into a virtual audio output.
linux, bash
curl -fsSL https://devfrp.github.io/SimulSpeaker/install.sh | bashInstalled: no root, nothing outside your home
$ speaker-sim --set-speaker "Genelec 8030C" --start
[ 50 %] Haut-parleur L : sons directs et réflexions
[100 %] Filtres prêts
$ wpctl status | grep Speaker
* Speaker Sim (Genelec 8030C) [Audio/Sink]
A top-down plan of your virtual room: drag yourself or the speakers, dial in acoustic treatment and materials, and watch the reflections and RT60 change live.
Every curve comes from published measurements, pinned to a fixed Git commit and downloaded once.
Five stages are baked into four minimum-phase impulse responses (left/right speaker to left/right ear), then run by PipeWire’s convolver in a separate, user-level process.
Listening Window for the direct sound, Early Reflections and Sound Power for what the room returns.
CEA-2034, spinorama
Real materials and acoustic treatment, absorption per octave band, RT60 per band (Eyring), and image sources filtered by every wall they hit.
RT60 / octave, Eyring
Each arrival is rendered through a KEMAR HRTF from its true direction, with exact fractional delays and diffuse-field EQ.
libmysofa, SOFA
Your model’s AutoEq parametric EQ, turned into a minimum-phase impulse response.
AutoEq
A virtual sink becomes your default output; stopping it restores the previous one, even if systemd stops it.
PipeWire, systemd --user
IR[s][e] = HP⁻¹ × ( DF⁻¹ × ( LW × Direct[s][e] + ER × Reflections[s][e] ) + SP × Reverb[s][e] )Live simulation needs PipeWire, so it is Linux-only. On Windows and macOS the same engine exports the filters for a system-wide convolver.
One command. Installs into your home folder, adds a menu entry and the speaker-sim command.
curl -fsSL https://devfrp.github.io/SimulSpeaker/install.sh | bashA single .exe. Builds the filters and writes a ready-to-include Equalizer APO configuration.
A native .app in a .dmg. Exports true-stereo impulse responses for any convolution plug-in or host.
Builds are produced by GitHub Actions from the tagged source; each binary is smoke-tested by actually exporting filters before release. SHA-256 checksums are attached to every release.
| Feature | Linux | Windows | macOS |
|---|---|---|---|
| Live virtual output | ✓ PipeWire | via Equalizer APO | via a convolver |
| Headphone + speaker selection | ✓ | ✓ | ✓ |
| Room simulation & curves | ✓ | ✓ | ✓ |
| Auto-start, output restore | ✓ systemd | no | no |
| Export WAV / Equalizer APO | ✓ | ✓ | ✓ |
Short answers. The README has the long ones.
Settings and filters in ~/.local/state/devfrp-speaker-sim, measurements in ~/.cache/devfrp-speaker-sim. Enabling auto-start asks systemd to link the unit, which creates symlinks under ~/.config/systemd/user; disabling removes them.
It is the share of the walls covered with absorbing panels. Panels soak up the mids and highs much more than the bass, so the RT60 meters drop above 250 Hz first; add bass traps to tame the low end. Six presets go from a bare room to a very dead control room.
Filters are normalised for the worst case (the same signal on both speakers), so nothing ever clips. Raise the preamp if your material leaves headroom.
Yes, for anything already in the cache. Sources are pinned to a Git commit, so a cached file never changes and is never downloaded twice.
It is the reference most measurements are built around. Any SOFA file can be set as sofa_file in the configuration if you have your own HRTF.
Re-run the curl command to update. Append “bash -s -- --uninstall” to remove the app (settings kept) or “--purge” to remove settings too.
AutoEq by Jaakko Pasanen (MIT), Spinorama by Pierre Aubert (GPL-3.0) with ASR Klippel NFS measurements, and the MIT Media Lab KEMAR set. Data is fetched on demand, not redistributed.