Image To Sound

This code converts an image file to a sound by treating the image as a discrete time fourier transform and inverting it.

Usage

install the requirements from requirements.txt and then run image_to_sound.py

./image_to_sound.py <image_file> | --fs <Hz>  | --out_duration <seconds> | --rescale <positive_float> | --out_file <name>.wav

The lower fs is the longer the output will be so use --out_duration if you need a low fs. --rescale raises the image intensity to the given power. Use a value > 1 to incrase contrast in dark images (most images I've tried look best with this). Use a value < 1 to increase contrast in bright regions.

To superimpose image's audio on other audio

$ ./image_to_sound.py image.jpg --fs 44100 --out_duration 298.06 # duration of song in seconds
$ sox -m -v 0.01 image_sound_file.wav -v 1 song.wav song_with_image.wav

Example commands

Let's play with the demo image.

$ # 1
$ ./image_to_sound.py demo/demo.jpg --out_duration 5
$ # 2
$ ./image_to_sound.py demo/demo.jpg --out_duration 5 --rescale 7.3
$ # 3
$ ./image_to_sound.py demo/demo.jpg --out_duration 5 --rescale 1 --out_file my_sound.wav

command 3 shows the importance of rescaling (default 10) with most color images.

Demo Image

Command 1

Command 2

Command 3