Hi,

This is my first post here! I've been lurking around a bit occasionally. There seems to be a lot of nice and smart people around.

I'm a complete newbie when it comes to Pd, but I've got my head around the basics. You might call me lazy for asking this (instead of wading throught he documentation), but I just thought I might get advice on a few different approaches to a patch that would:

1. Create a list of sub-directories that recide in a 'main' directory
2. Randomly pick one of these directories
3. Randomly pick one of several MP3 files in this sub-directory
4. Randomly pick a start time (offset) in that file
5. Randomly generate a sample time (say, 2-20 seconds) of data which is extracted from the file, X amount of time into the file (as determined by the previous step). And also making sure that the file is long enough for this to happen (that the sample length does not exceed what's left of the file after X seconds of 'offset').
6. Generate a fade-in/fade-out (envelope?) for this sample
7. Place it somewhere in the stereo field (panning left or right) or even better, a placement in a 5.1 surround field.
8. Play it
9. Repeat and layer lots of random sounds from random locations according to a set 'density' control in the patch.

And also, how would I set up Pd to output 16 channels into Soundflower for recording individual tracks and later mixing them surround?

Regards

enumerating the directories and picking a random file is quite easy
using [shell]. You could use find or locate to do this, locate will be faster.

The old [playmp3~] is wobbly, I recommend you try [readanysf~]. I'm not sure about
seeking to a random location in the file though, you may find this is easier with
.wav files than .m3s

Is there a simple way of reading this data (2 to 20 seconds) from within an mp3 file into, say, an array? Or maybe loading the whole mp3 file into an array, and then numerically measuring how long it is, extracting randomly a piece of it and then unloading the part of the file not needed?