179,854Messages
9,130Senders
30Years
342mboxes

← back to listing · view thread

From:
Kent williams
To:
Anig Browl
Cc:
IDM List
Date:
Wed, 29 Aug 2001 13:37:23 -0500 (CDT)
Subject:
Re: [idm] Dissecting Windowlicker track #2
Msg-Id:
<Pine.HPP.3.96.1010829132439.27122B-100000@arthur.avalon.net>
In-Reply-To:
<013901c130b7$43060160$7ab8869f@pauls>
Mbox:
idm.0108.gz
The big problem with any of these picture to audio convertors is that they're limited to cartesian frequency VS amplitude plots. It's similar to the problems you run into with compressing image data -- a row/column traversal doesn't match the 'importance' weight of a picture. A prof of mine was a compression head and had some success with improving image coherence by using hilbert space-filling curves to traverse an image's pixels. That seems to improve the chance of exploiting neighbor-similarity. What I would do in a Coagula-type program to make it reflect a picture's composition I'm not sure, but some alternative mappings could be interesting as a way to give different results from the same image. I haven't seen anything like this on PC yet, but there's also some cool programs for SGI and other Unix boxes that allow you to diddle the sonogram in interesting ways, producing spectral shifts and compressions. If I'm not mistaken, you can use Coagula to build sonogram images from audio files too. It's interesting to take a sonogram of a sound and manipulate it in a photo editor, then re-render it. Simple transformations produce very interesting results. Horizontal stretching == time expansion, and pallette manipulations can screw up a sound radically. You can also take two sonograms, and do a visual crossfade of them to get a spectral sound morph. What fun. Too bad I can't blow off work for the rest of the day and go play with this stuff... --------------------------------------------------------------------- To unsubscribe, e-mail: idm-unsubscribe@hyperreal.org For additional commands, e-mail: idm-help@hyperreal.org