chopping 16's

yo mpc fam, i don't always chop in this fashion, but sometimes, i like chopping a loop in 16's so i can replay the sample. i.e.) chop 16 for 4 bars of 4/4 so each pad plays a 1/4 note of the sample. it seems like 9th wonder does this for all his beats (sort of annoying) and pete rock does this too for one of his vids : http://www.youtube.com/watch?v=z73CcodfT_w


my question is, regarding tempo. i have my unchopped 4 bars of 4/4 perfectly looping at 92bpm, but when i chop it into equal 16 and quantize on the 1/8th and record it in a new sequence at the same 92bpm, there are little spaces and "clicks" in between each beat. i understand that drummers and musicians don't play exactly in perfect timing, but shouldn't the unchopped sample loop at 92 be identical to the chopped sample loop at 92? shouldn't they be played the same thing at 92?

because of this, i always pitch down my sample to get rid of the gaps, but then the metronome speed is too fast and the sample is too in the back of the beat so my drums feel mad rushed.



good looks!

I think this is due to the mpc setting anti-click envelopes at the start/end of samples.. it very quickly ramps the volume up and down when starting and stopping the sample, to prevent clicks, but it also seems to prevent proper back-to-back playback like you describe... I don't know for sure but its a good guess.
Have you experimented with release times on the chops, and setting them to poly mode (so each is slightly long, and they dont cut one another off. This allows some overlap and may help).

leonard-lee wrote: i understand that drummers and musicians don't play exactly in perfect timing, but shouldn't the unchopped sample loop at 92 be identical to the chopped sample loop at 92? shouldn't they be played the same thing at 92?

I would say yes they should be, and no, they are not actually played 'properly'. If you want to eliminate the drummer, create your own perfectly quantized hits, and have some pad sound fill out the length. The problem you mention is still apparent (its also there when trying to mix down a 1bar loop, and then having it trigger at the start of the bar.. wont loop smoothly).

Add a bit of decay to all your chops so you don't have a click at the end. What i also do is making each chop a bit longer than it should be: you leave a bit of the unwanted sound at the end of your chop, and then you use mute groups so each pad mutes the one that was played before. It works almost everytime.

Not sure what mpc you are using but if you chop by zone on the 2kxl you can set a margin which pretty much adds a few milliseconds of the start of the next chop so its blends in more smoothly. If not try time stretching because as you noted musiciens dont play quantized and some phrases will be shorter some longer. You can also change the tempo a little bit instead of pitching down.

Just add attack and decay to the chops.
Like 5-10 attack and your chops will run smoothly

what you need to learn about is "loop mode," you can eliminate all those clicks and even play with the bpm a bit all while the sample loops fluid.

i break it down here on the last post of this thread. it was one of my major "a-ha" moments in beat making.

viewtopic.php?f=26&t=163030

yeah, the reason you are getting the clicks is because you may have not chopped exactly on the bpm, or the sample may be drifting in and out of time slightly, it happens with all samples pretty much, unless its an electronic one thats snapped to a grid, then again you may not be chopping every sample equally, and if you are it still may not fit because of the 96 ppq of the mpc. (i hadnt thought of the ppq thing before, but it sounds right to me.)

as a sample based producer youll run into this kind of stuff all the time, the key is to develop workarounds to make it all gel together even if its not exact.

I think the point is that, even if it IS all exact, the MPC will not play it back smoothly due to anticlick envelopes or that it is not sample-accurate playback. So wat you are actually doing with your attack/release techniques there is covering up for issues in the MPC engine.

Generate a sample that contains an exact 441hz sine tone (must be sine wave!!). Cut it to exactly 44100 samples long. so it will be exactly one second (100 cycles of the wave). Next, set the bpm to exactly 240. Turn off all attack and release values. Trigger this sample at the exact loop start (use step edit or quantize to place it exactly on the 1).

The result when looped should loop EXACTLY around, right on the 100th zero-crossing point. Does it?

If you think the 96ppq is a factor (could well be, see below) then repeat the experiment but use an 8 second wav (352800 samples) and a bpm of 30. This should then loop EXACTLY on a zero crossing, on exactly one of the 96ppq events.

The 1 second example loops around between two of the PPQ ticks, so the root problem might be that the second time the sample is played, it is played late (on the following ppq tick), which creates a little gap. If you record the output of the 1second experiment and study the gaps, you should see it widen and close and finally match up on the 8th loop...more or less

I did the maths bit, anyone else want to do the practical work?