A Python implementation of MIT's Visual Microphone [1].
Our code is based on the original MATLAB version. It depends on OpenCV for handling videos, SciPy for signal processing and writing audio files and pyrtools [2] to compute complex steerable pyramids. Other requirements are NumPy and Matplotlib.
python video2sound.py [-h] [-o OUTPUT] [-s SAMPLING_RATE] input_video
positional arguments:
input_video The path of the input video
optional arguments:
-h, --help Show this help message and exit
-o OUTPUT, --output OUTPUT
The path of the output file
-s SAMPLING_RATE, --sampling-rate SAMPLING_RATE
The frame rate of the input video
Antonio Musolino and Davide Sforza.
[1] DAVIS, Abe, et al. The visual microphone: Passive recovery of sound from video. 2014.
[2] PORTILLA, Javier; SIMONCELLI, Eero P. A parametric texture model based on joint statistics of complex wavelet coefficients. International journal of computer vision, 2000, 40.1: 49-70.