I really like NightRadio's work. His main synthesizer, SunVox [1] is amazing. Innovative UI, very good analog-modeling, and a wide assortment of synthesis features. Oh, and it is available on almost any platform you want, most for free.
Also, I suggest checking out the NightRadio's other project, PixiVisor. [2] It converts an image (progressive scan) to audio and back again. This lets you, for example, see an image processed by a reverb effect, resonant LPF or whatever. It does things with the Fourier transform that I would have guessed were impossible.
Then just add some standard chorus+reverb processing, possibly a bit of distortion.
There are other programs that do this kind of thing (Photosounder, MetaSynth), oddly not so much free software.
As a side note, it's kind of odd that the coloring in this gui lays out the light spectrum with the low frequencies in violet and the high frequencies in red.
The Analysis & Resynthesis Sound Spectrograph is the closest thing I can think of: http://arss.sourceforge.net It hasn't been updated in a long time but it's still very useful!
Michael Klingbeil's SPEAR is also open source, but a slightly different beast - focused on manipulation of analysis data as a set of partials - it's still insanely useful: http://www.klingbeil.com/spear/
EDIT: Just to say that Alexander Zolotov's work is blowing me away. I didn't realize he was behind SunVox until now, but the portfolio of his projects in this thread is really inspiring.
Also, I suggest checking out the NightRadio's other project, PixiVisor. [2] It converts an image (progressive scan) to audio and back again. This lets you, for example, see an image processed by a reverb effect, resonant LPF or whatever. It does things with the Fourier transform that I would have guessed were impossible.
[1] http://warmplace.ru/soft/sunvox/
[2] http://warmplace.ru/soft/pixivisor/