Copyright (C) 2010-2020 Carl Schissler, University of North Carolina at Chapel Hill. All rights reserved.
GSound is a physically-based sound propagation package used for acoustic simulations in various environments, developed by Dr Carl Schissler. pygsound is the Python package that wraps GSound's codebase for efficiently computing room impulse responses (RIRs) with specular and diffuse reflections. GSound is powerful enough to be used for sound simulation in 3D scenes with complicated geometry and acoustic materials. This repo's python API has not exposed all of GSound's components. But we do provide the complete C++ source code and welcome pull requests if you made useful modifications (mainly the python API).
This repo has been configured to build with CMake (version>=12), and mainly tested on Linux and MacOS.
On Linux, install dependencies using:
sudo apt-get update
sudo apt-get -y install cmake python3-dev gobjc++ libfftw3-dev
On MacOS, install dependencies using:
brew update
brew install cmake python3 fftw pkgconfig
Note: if you use a virtual environment, please make sure the python version in your virtual env is consistent with your system-wide python version. Otherwise the python3-dev package may not be found by the CMake and is unusable.
First clone this repo with:
git clone --recurse-submodules https://github.com/RoyJames/pygsound.git
Then you can build and test with
cd pygsound
python3 setup.py develop
python3 setup.py test
or directly install it as a python package with
cd pygsound
pip3 install .
Note that the test module requires libsndfile, which is an additional dependency you need to install.
See examples folder. You need to cd examples and run python3 mesh_sim.py (we recommend starting with this one). This script demonstrates two equivalent ways to define the environment for sound propagation, and save the impulse response as an audio file. You can use a .obj file with an optional .mtl file with the same name to define the room geometry and materials. In this case, the .mtl file has two extra rows compared with conventional .mtl file used for visual rendering:
sound_a 0.5 0.6 0.6 0.7 0.75 0.8 0.9 0.9 # sound absorption coefficients, for 8 octave bands [62.5, 125, 250, 500, 1000, 2000, 4000, 8000]Hz
sound_s 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 # sound scattering coefficients, if you don't know the details of diffuse/specular reflections, keep it 0.5
or directly create a shoebox shaped room using our API:
room = ps.createbox(dim_x, dim_y, dim_z, absorption_coefficient, scattering_coefficient)
The benefit of using the .obj style is that you can easily define different reflection/absorption coefficients for each triangle element for each frequency sub-band.
This sound propagation engine has been used for many research work of Dr Carl Schissler and other researchers in the UMD GAMMA group for audio rendering and impulse response generation purposes. For example:
@inproceedings{schissler2011gsound,
title={Gsound: Interactive sound propagation for games},
author={Schissler, Carl and Manocha, Dinesh},
booktitle={Audio Engineering Society Conference: 41st International Conference: Audio for Games},
year={2011},
organization={Audio Engineering Society}
}
@article{schissler2017interactive,
title={Interactive sound propagation and rendering for large multi-source scenes},
author={Schissler, Carl and Manocha, Dinesh},
journal={ACM Transactions on Graphics (TOG)},
volume={36},
number={1},
pages={2},
year={2017},
publisher={ACM}
}
@inproceedings{9052932,
author={Z. {Tang} and L. {Chen} and B. {Wu} and D. {Yu} and D. {Manocha}},
booktitle={ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
title={Improving Reverberant Speech Training Using Diffuse Acoustic Simulation},
year={2020},
volume={},
number={},
pages={6969-6973},
}
For a complete list of relevant work you may want to cite depending on how you use this repo, see our speech related research and sound related research.