ModalSound Source Code and Dataset

Source Code

Source code can be downloaded here. Compilation instructions are included in each folder in both html and markdown format. We have compiled and tested our codebase on Ubuntu and macOS machines.

Dataset

We validate our code using a benchmark dataset Thingi10K. We pre-process all 10,000 meshes such that the remaining meshes satisfy the following: After the pre-processing, we have 4968 meshes. We then run IsoStuffer to tetrahedralize all the obj mesh files. The resulting *.tet files are ready to be used as input to ModalSound pipeline.

All the input/output files can be downloaded (input obj (1.5GB), output tet (16GB)).

On this webpage, we visualize all the meshes.

Contact

Dingzeyu Li (dli@cs.columbia.edu) and Changxi Zheng (cxz@cs.columbia.edu)