I've been working on a simple looper (no overdub). I have almost all the midi control and effects I want, but the timing on the loop just doesn't seem to line up.

The patch just records a loop(tempo synced to the main metronome) using tabwrite~ to an array. Then, when the end of the user determined loop length(in measures) is reached, tabplay~ is triggered and the lop plays back, every time the loop length is reached tabplay~ is retriggered.

Anybody have a more elegant solution or insight into why the timing is messed up. any help would be appreciated! i'm using an mbox with very low latency so that isn't the issue either

i'm attaching my file, the guts of the loop code can be found under [pd audio], [pd loop1], [pd loop1a] and [pd loop1b] from the main page (all the controls are set up for loops 1-8(a and but only loop1 has active code so far, i want to get it just right before copying it 7 more times.

tabplay~ and tabwrite~ only start on block boundaries. which is more than a bit annoying. partial solution for tight playback timing: build your own tabplay~ with vline~ and tabread~ or tabread4~, not sure if there is a vtabwrite~ anywhere though

the problem is that [metro] is more accurate than [tabplay~]/[tabwrite~] - metro is 100% bang on time every time, but tabplay~/tabwrite~ only start on block boundaries (which is usually 64 samples, around 1.5 ms).

Thanks, I switched to a vline~ driven ~tabread4 and a user defined delay compensation before the tabwrite~ and the timing has improved dramatically. I also added a much more accurate timebase found from old posts on this site.

What do you mean by a user defined delay compensation? Won't the inaccuracy be different every time because it depends on when within the block you stop recording?