Has anyone of you ever heard about a patch that converts wav or aif file to bitmap, so that you can edit it visually by using photoshop or any other graphic editing software? I have just seen that is done by using Adobe Audition, and the guy who showed it to me said that he remembers someone tried this with Pd...

Yeah, you could do this in gem. Was the picture a standard visualization; Time/Amplitude, Frequency spectra or what?

I don't know how to do it in pd, but I've got some luck using ARSE (arse.sourceforge.net). It converts the audio into a spectral bitmap, and then converts it back.

@nestor said:

Yeah, you could do this in gem. Was the picture a standard visualization; Time/Amplitude, Frequency spectra or what?

I didn't get a chance to have a closer look but as far as I can see the audio was represented in two different images; one for frequency domain, the other for phase, exactly as you said... How on earth can something like that be done by GEM? Is it possible to save the output of GEM as image file to open with a different application? It seems to me like a new dimension in audio editing and synthesizing, and if there is any way to get my hands on it other then buying an expensive software, I would definitely go for it nestor!

[pix_record] or something like that, isn't it?

As far as I can understand, [pix_record] outputs as a movie file. What I am looking for is to have the output as bmp (bitmap) or some other format like jpeg, tif, etc., so that I can open it with GIMP, photoshop, or any other graphic editing software.

What I saw that friend of mine did was, he was opening the converted file, applying various filters, like blur, clouds, etc, then convert it to audio file again, which has been manipulated according to what he did in photoshop; i.e. applying blur filter makes the source sounding in a muddy way, or the shapes he draws makes it distorted, and so on...

Well, I think you could do this in GEM.... First, you decide what you want to make your picture out of... let's say the real and imaginary part of the fft. So, dump these into two arrays with [tabsend~]. This would give you one block (64 samples, default) of data for each part. Then, you could convert this to a set of images using [pix_set], ending up with a greyscale image 8*8px. Texture this to a square of the appropriate size, or not if you want to add aliasing, then [pix_record] to dump a .jpg or whatever.

Then to do the inverse, load a pix and then use [pix_dump] to get all the data back in a table and read it with [tabread~] and then finally resynthesize.

Of course, you could do all this in subpatches and use [block~] to change the blocksize to get a more interesting (higher res) picture.