A sonogram (also called spectrogram) represents graphically the distribution of energy of a sound through frequencies, in function of time.
It may be funny or useful to transform the data graphically before resynthesizing the sound. This is made easy by a development environment like Max/MSP/Jitter, which allows to work in real time with sound and video.
In this video, I show a few graphical treatments one can apply to a sonogram:
If you wish to realize similar sound processing, here are Max/MSP/Jitter patches for spectral sound processing.
Update: Computer Music Journal Tutorial on Graphical Spectral Sound Synthesis
Read more explanations about this type of graphical sound synthesis in the free pdf article: A Tutorial on Spectral Sound Processing with Max/MSP & Jitter.





