LCC - An open-source version of our spatial audio algorithm for binaural playback over closely-spaced stereo loudspeakers.

Developed by Robert LiKamWa (@roblkw_asu) and Matthew Lane (@mattlane66), with guidance from Ralph Glasgal (https://www.ambiophonics.org/).

LCC (Localization Cue Correction) is a solution for spatialized audio through stereo speakers. LCC is a lightweight implementation of crosstalk cancellation so your left ear hears the left channel of audio, and your right ear hears the right channel of audio. LCC uses crosstalk cancellation to purify stereo imaging, emulating a physical barrier positioned between left and right ears and speakers. This digital barrier (LCC) preserves spatialization perceptual cues (ITD and ILD) and reduces undesirable anomalies, which are found in nearly all loudspeaker playback environments.

LCC works with symmetric head placement with respect to the speakers in front of the speakers. That is, the speakers should be arranged in a narrow/skinny isosceles triangle configuration, not an equilateral triangle configuration (sometimes called the 60° stereo triangle). Please read our white papers found in the link under the Theory section of this page for more detailed information and configuration suggestions to get the best possible results. Hear low-quality previews here and here. Just make sure your stereo loudspeakers are closely spaced.

Setup (macOS)

Install virtual audio device so that you can route audio inputs and outputs (e.g., https://github.com/mattingalls/Soundflower/releases/ found in the assists sections).
In your Apple System Preferences for Sound, set your desired speakers to play at a comfortably loud volume. You can further adjust the system volume later with your virtual audio device / Soundflower.
In the Apple System Preferences, select the virtual audio device (Soundflower) as your default audio output device.

Setup (Windows)

Install virtual audio device so that you can route audio inputs and outputs (e.g., https://www.vb-audio.com/Cable/index.htm).
Select your virtual audio device / CABLE Input as your playback device.

Setup (Linux)

Using pulseaudio

Make a virtual audio device , called a sink, with pacmd

a. Make a null sink to act as a virtual cable between audio input and speaker output

$ pacmd load-module module-null-sink sink_name="Game"

b. Look for device description under the name Game of virtual audio device in sinks from output of pacmd list sinks.

$ pacmd list sinks

You should see something like this which is what lcc will also show for querying input device. x sinks available. ... name: <Game> ... device.description = "Null output"
After running lcc and connecting it to sink Game, ie. "Monitor of Null Output", as input and stereo speakers as output, open PulseAudioVolumeControl

$ pavucontrol
Check that the RtAudio is connected to stereo speaker output in Playback tab and RtAudio is connected to "Monitor of Null output" in Recording tab.
The audio stream of the program in playback tab should be set to the device description "Null output" in order to route audio input to the sink Game which is connected to the speakers.

Running LCC

Double-click on lcc_audio on macOS or lcc_audio.exe on Windows to run it. You may have to right-click it and open it to grant permissions to run the application. You may also need to open the terminal before opening the lcc_audio app on macOS and run chmod +x PathToFile (in this case, the PathToFile being lcc_audio).
When LCC asks you to Select input device:, type the numerical input value for your new audio device and press enter.
- If you're using Soundflower, this will read ingalls for Cycling ’74: Soundflower (2ch).
- If you're using VB Cable, this will read CABLE Output (VB-Audio Virtual Cable).
- If you are using pulseaudio, this will read Null output.
When LCC asks you to Select output device:, type the numerical value for the desired speakers. (For example, Apple Inc.: Built-in Output for your MacBook's internal speakers.)
Set your preferred sample rate, e.g., 44100.
Press 1 and then enter to turn on LCC. Press 2 and then enter to turn off LCC.
See below for advanced parameter tuning.

Compiling code (macOS)

To compile the code on a Mac, navigate to the directory, open the Terminal, and run:

g++ -lpthread -framework CoreAudio -Wall -D__MACOSX_CORE__ -framework CoreFoundation lcc_rtaudio.cpp RtAudio.cpp  -lm -o lcc_audio;

Compiling code (Linux)

Navigate to the directory, and use the following commands. This will use cmake to build the lcc_audio binary. To compile the code on a linux or unix distribution, open a terminal, and run:

mkdir build
cd build
cmake ..
make
sudo make install
cd /usr/local/bin
sudo chmod +x lcc_audio

Alternatively, directly compile the program if cmake doesn't work.

g++ -O2 -march=native -mtune=native -DDATAPATH=$HOME/lcc_audio/data/ -D__LINUX_PULSE__ -D__LINUX_ALSA__ -D__UNIX_JACK__ lcc_rtaudio.cpp RtAudio.cpp -o lcc -pthread -ljack -lasound -lpulse-simple

Compiling code (Windows, MinGW)

You need to have MinGW installed:

choco install mingw

To compile the code on a Windows, navigate to the directory, open the PowerShell, and run:

g++ -Wall -D__WINDOWS_WASAPI__ .\lcc_rtaudio.cpp .\RtAudio.cpp -static-libstdc++ -static-libgcc -lole32 -loleaut32 -lmfplat -lmfuuid -lwmcodecdspuuid -lksuser -lm -static -o ./lcc_audio.exe

LCC and Scripting

To implement scripting with LCC, 8 arguments must be entered in the command-line to active the scripting mode which is just calling function setupAudioStreamNoConsoleQuery().

The function setupAudioStreamNoConsoleQuery() reads from text files for its input in order to keep the original functionality of lcc running a loop and checking for input.

Check documentation/scripting-with-lcc.md for details of how the process works and how to implement your favorite scripting language to use lcc.

Bash script sample provided in lcc-script-interface/bash/bash-script-sample.sh.

C++ files for interacting with lcc provided in lcc-script-interface/cpp/ directory.

Theory

https://medium.com/@mattlane66/spatial-audio-for-speaker-connected-devices-cdf96d127f67?source=friends_link&sk=6c8494ea1eec58bd1458e156a8fa6418

Forum

https://news.ycombinator.com/item?id=23649673

Attribution

This is free and unencumbered software released into the public domain.

Anyone is free to copy, modify, publish, use, compile, sell, or distribute this software, either in source code form or as a compiled binary, for any purpose, commercial or non-commercial, and by any means.

In jurisdictions that recognize copyright laws, the author or authors of this software dedicate any and all copyright interest in the software to the public domain. We make this dedication for the benefit of the public at large and to the detriment of our heirs and successors. We intend this dedication to be an overt act of relinquishment in perpetuity of all present and future rights to this software under copyright law.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

We ask that if you use this software for anything that you please attribute to authors Robert LiKamWa (@roblkw_asu) and Matthew Lane (@mattlane66) at L&L Innovation Labs.

MeteorStudioASU / lcc