Structure from Sound – python

Setup

Prerequisites

To use this repo, you need to have MATLAB and Python installed.

Download Data

• Download data from here and put the files in the data folder,

├── data
│   ├── tdoa_20201016
│   │   ├── data
│   │   │   ├── chirp_0001
│   │   │   │   ├── Track 1.wav
│   │   │   │   ├── Track 2.wav
│   │   │   │   ├── ...
│   │   │   ├── music_0014
│   │   │   │   ├── Track 1.wav
│   │   │   │   ├── Track 2.wav
│   │   │   │   ├── ...

Matlab

Run matlab/setup.m

Copy config.json.dist and rename the copy config.json. Then edit config.json to include the path to your matlab installation.

Python

Install python dependencies using pip3 install -r requirements.txt

Using GUI-plot

In order to use gui_plot.py first run genereate_detection_data.py (takes ~10 min). Then run gui_plot.py

Structure sketch

Data Formats

• Audio is stored as .wav files where each experiment is stored in a serperate folder with microphones i recording stored as Track i.wav
• TDoA-matrices (i.e. detections) are stored as <experiment name>.npy with the stored array being a tensor with shape #detection_windows, #mics, #mics where the value stored in position (t,i,j) is |s(t) - r_i| - |s(t) - r_j| (units meters) (TODO: double check sign of this)
• Positions, to be decided...