Hi David,
quick late night idea: write to two tables and overlap them with a crossfade by 5-10 ms or so. I think your time-triggered ramps at the start and end points of the loop are still a good idea anyhow. If you are doing any kind of resynthesis of the recorded sounds, you might be interested in taking a look at this live looping & granulating abstraction I posted a while back:
http://lists.puredata.info/pipermail/pd-list/2006-02/035619.html
best! d.
David Schaffer wrote:
Hi,
I'm building a little time domain abstraction that captures audiofrom an input and writes it on the fly to an array, this array is then beeing red continuously as a loop (as a result, the incoming sound seems to "freeze", see what I mean?). my problem is avoiding clicks at both ends of the loop; right now, I'm trying to use time based triggers to create a "windowing" function so that the amplitude of the loop is zero at both ends, butI was wondering if there was a way to tell pd to start and stop it's recording process at zero crossing points, or to truncate its reading points a bit so that the beginnig and the end of the loop are zero. Can anyone help me in this?!