[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Date Index][Thread Index][Author Index]

Re: dream box



[Sometimes I feel that, if I'm into this "looper religion" thing, 
then Kyma is like my denomination. :)  That makes the MAX/MSP folks 
(and Orville users, etc.) like a different denomination; same 
religion, they just use different words to mean about the same thing!]

Yeah, there used to be a sign up in the maintenance closet at a 
studio I worked in: One World One Love One Piece of Gear.


>
>(Brainstorming here, haven't completely thought this out) Seems like 
>the trick might be to record it initially without clicks. In other 
>words, clean it up on the way into the loop memory instead of on the 
>way out (at playback).  Sort of like a "smart overdub".  It would 
>delay the recording by a few sample points until an appropriate 
>zero-crossing, then delay the "punch-out" until it saw a matching 
>zero-crossing (indicated by the first derivative).  You might need 
>do double-buffer the baby.  I.e., put the overdub into a working 
>buffer, trim it neatly, then shuflle it into the main loop memory.

Yeah I think we're on the same track more or less...but let's say you 
wanted to lay in a thick drone in a 10 sec. sub-loop. While laying 
that in, I'd want to hear a seamless loop just the same as if the 
total loop length was 10 sec. No problem there with the incoming 
audio at least. But then when I "zoom out" I'd want to hear that 10 
seconds in its entirety, but the ends are now "unrolled", so the 
drone would start very suddenly in the context of the master loop. So 
I was thinking of copying a little bit of the end of the subloop to 
before its beginning, and vice versa, with adjustable fades in and 
out.

And vice versa, assuming there was pre-existing sound at the 
boundaries of the 10 second segment, when I "zoom in" and start 
sub-looping, the pre-existing sound might click but that's where what 
you describe should work just fine. Or, if the new material is going 
into a working buffer, the master loop might use an overlap-crossfade 
playback method instead of searching for zero-crossings, I'm not sure 
which would be better. This is about where I usually get a headache.

I hadn't resorted to double-buffering in previous attempts, but now 
it seems likely it's the way to go. Thanks for the collaborative 
brainstorming!

-Alex