Skip to content
Switch branches/tags
Go to file

Latest commit


Git stats


Failed to load latest commit information.

This was an attempt at creating a sound from an image of my cat. The idea was that, since a spectrogram is an "image" representation of a sound, an arbitrary image could be interpreted as a spectrogram of a sound. With that assumption, I attempted to create a sound that would have the spectrogram of an image of my cat.

I quickly realized that starting with an image of my cat was a bad idea, because I had no way of telling if the sound I created was correct or not. So, I decided to start with an actual spectrogram of an actual sound, and then try to reconstruct the original sound from the spectrogram. Then I could tell if I was even close or not.

Later I found that this is actually a "longstanding problem in audio signal processing" and is anything but trivial. So, needless to say, I never "completed" this project. But I think I'm finally okay with that.

Fair warning: you do not want to listen to any audio files that this program creates. They are wildly unpleasant.

I really need to work on organizing this repository. In the meantime, pay most attention to and additive-synthesis.ipynb.


  • libjpeg
  • libsndfile
  • pil
  • scikits.audiolab
  • matplotlib
  • numpy
  • scipy


Making sounds out of images of cats.



No releases published


No packages published