I've got this idea but haven't began working on it so I'll ask here if anyone has some comments or suggestions. Thanks in advance
I've built an installation using pd, that basically detects (voice) input, waits and then throws back a pre-recorded answer. - It's all done very simple(envelope follower, delay and random selection of samples) but in practise it works surprisingly well.
What I would like to do is to record the input, mostly its just none, so record it only when somebody actually speaks/uses the installation.
This is the part I would like some help with, should I have a relatively short buffer(or two alternating perhaps) that I record into and if nothing happens just drop it and begin over again - (until I actually record *something*)?
Then, how can I trim the sample if I manage to record it? I could have like half a second or less silence at the beginning and cut it about the same time after the amplitude has dropped - how should I build such a thing?
I anybody could shed some light or point me in any direction it would be very helpful.