the calculation as well, so that you could deliberately stagger the blocks and more evenly distribute the calculation in cpu-intensive situations. I'm imagining something like two 4096 blocks running say, 64 samples apart so that one is does its calculation while the other is still collecting samples.
Matt
Maybe a master to sequence the starting of the blocks. instead of using block use switch~. If the masters blocksize is 64 set up a counter that sends a bang every iteration trigger the first switch at 4096. next iteration trigger the second. do a mod 64 repeat the process.
I had no idea that convolution and fft vocoder were that much different. I will have to look up convolution now my only experience with it was in cool edit pro. After hearing Aphex Twin's Bucephalis Bouncing Ball I recorded a ball bouncing, extracted the timing in cakewalk, made some drum sounds trigger at the ball bounce, imported to cool edit, then convoluted with different sounds like female voice. Then there was Sonic Foundry's Acoustic mirror which sounded really muddy. Actually my use of the fft vocoder could almost be done with an envelope follower.