Experimenting with sound spectral analysis via spectrogram ( a visual representation of the spectrum of frequencies in a sound or other signal as they vary with time). I found out that is possible to reverse the procedure and generate sound frenquencies by image input analyzed and transfered through spectogram.

I’m creating this surreal endless block of patterns/structures found in local urban areas of Iasi, Romania, describing the urban eviroment and living in the town. I’ve picked several close-up pictures which works as a series rather than self-describing single ones, and one following the other invokes the impression of characteristic repetetive moments commonly found in local surroundings. Particual pictures were chosen and ordered with knowledge of sound outcome in the mind. Knowing which shapes and patterns will cause particular kind of audio result. So the whole image stripe was constructed and treated like an composition.

I’ve tried multiple softwares to convert the image to sound. There are various ways the spectogram can analyze the image and generate the sound. In this case it works this way: The vertical range defines frequency range from lowest at the bottom to highest at the top. The horizontal determines the time of the playback. All values like frequency range or speed of the playback are definable. I work just with monochrome images, so black stands for silence and white for maximum volume value. The image serving as the input has been edited in terms of contrast or color invertion to achieve the desired effect. No additional audio postediting has been done except continuous volume control.

Presented work is done in prety simple software called "Audio Paint", which generates pure frequencies by number of osscilator given by height of the image in pixels, which I prefer over different ones which for example simultates some synthetizers etc.