Revisiting jitter/drift issues on my MPC

Hi,

I know this subject has been covered, but I thought I could provide some more information and open up some discussion.

tldr; if you are overdubbing, you have to do a one-pass of your whole song, not starting your mpc a bar before you punch in, even if MIDI synced.

This is why:

I tested my MPC2000XL timing by recording multiple takes of 5min with the click on. Some were MIDI synced via MTC, others not.
This was done with Pro Tools 2024.3 and an SSL12 audio interface with midi I/O at 24bit/48kHz. When synced, PT was the master.

The result was quite interesting:
- 120 BPM on the MPC were 119.9830 on Pro Tools. This kept them in sync for 5min.
- Within the 5 min, some clicks occur before the PT grid and some after. The maximum difference I found was 36 samples late and 16 samples early

This is the most important finding for me:
- All recordings matched to each other almost perfectly on every click, no matter if they were MIDI synced or not, as long as they were aligned on the 1st click.

Example:
Let's say I recorded 8 tracks of my drums and I want to replace the cymbal that comes on in the middle of the song.
- If i try to start the MPC a bar before the cymbal part, and just punch in that part, this will have a drift/jitter that will be different from the other tracks. This can introduce differences up to 1ms! so goodby to your "MPC groove" with that.
- The solution is to start the MPC at the same time the drums start, and let it play until the middle of the song where the cymbals come in. In this way, the MPC will be drifting/jittering in the same way as in the other tracks.

Another issue is when syncing PT and the MPC. Once PT starts playing, the MPC starts too, but there is a difference in timing between each sounds. Specially between the sounds coming out of the 8 outs vs the sounds on the master channel. This effectively renders the computer as useless for recording multi-track.

have you look into "delay compensation" ? or "latency" compensation.

What are your buffer settings as well

emanuellauria wrote:The result was quite interesting:
- 120 BPM on the MPC were 119.9830 on Pro Tools. This kept them in sync for 5min.
- Within the 5 min, some clicks occur before the PT grid and some after. The maximum difference I found was 36 samples late and 16 samples early
...
- If i try to start the MPC a bar before the cymbal part, and just punch in that part, this will have a drift/jitter that will be different from the other tracks. This can introduce differences up to 1ms! so goodby to your "MPC groove" with that. ...

nothing to worry about: you observed a range of 52 sample difference.
which represents almost exactly 1ms at 48khz sample rate

Telefunky wrote:
nothing to worry about: you observed a range of 52 sample difference.
which represents almost exactly 1ms at 48khz sample rate

can our ear even notice it ? at least mine doesn't

You can hear the difference:

Take a kick sample, duplicate it, delay the duplicate by 1ms. Play it. It will sound terrible.

Phase issues, comb filtering, prob loss of low end, smeared transient, etc. Modulate the delay from -1ms to +1ms and you'll have created a flanger effect.

I think we can't hear a difference below 0.5ms (at 48kHz that's half 24 samples )

chilom wrote:have you look into "delay compensation" ? or "latency" compensation.

What are your buffer settings as well

I have, that helps with the first click aligning to the grid better, but not to the drift/jitter that occurs afterwards, nor the difference in timing between each channel that I mentioned on the second post.

emanuellauria wrote:You can hear the difference:

Take a kick sample, duplicate it, delay the duplicate by 1ms. Play it. It will sound terrible.

Phase issues, comb filtering, prob loss of low end, smeared transient, etc. Modulate the delay from -1ms to +1ms and you'll have created a flanger effect.

I think we can't hear a difference below 0.5ms (at 48kHz that's half 24 samples )

not trying to argue here, but i would say that phase issues are a bit different than timing discrepancy "perception"..as they can be obvious even with extremely low latency no ?.

regarding your channels problem, are you using some FX or anything ?

chilom wrote:
not trying to argue here, but i would say that phase issues are a bit different than timing "perception"..as they can be obvious even with extremely low latency no ?.

regarding your channels problem, are you using some FX or anything ?

I can agree on that. Maybe I'm going crazy with this.

Regarding the channels, in my test, a set of hi-hats was playing off-sync, sort of delayed, while the rest of channels were on time. I realized that by setting the quantization to 50% and 0 (early/late), it plays fine. The problem arises only when I set "earlier" to some value. "Later" on the other hand, plays fine too. It doesn't matter if I play an empty sequence before it, it still fails to play the hi-hats on time if they have been moved "earlier".

I tried using my RolandVS as master and the problem was there too, but only when using MIDI Clock. When using MTC (Midi Time Code) the problem was resolved. Now I'm trying to get Pro Tools to play MTC, but for the love of life I can't figure it out.

Ok, figured out how to use MTC in PT and now the MPC syncs very well. Hi-hats are now in time, no matter what quantization setting they have. Also BPM is now 120, as set by PT. The MPC drifts by +/-1 ms as mentioned before, but as @chilom and others say, it's no biggie. Thanks!!

You can always do it the old way of not syncing.

Use song mode put a 4 click 1 bar sequence the same tempo as your beat. Place this before your sequence intro / beat you’re recording whatever. Let the whole thing play out in real time into the daw on however many channels you are tracking.

Ie a kick in the 1 2 3 4 quantized rigidly.

Record enable your daw hit play on song and the 1 bar sequence will be there. You then drag the whole batch of audio and line that kick up on your daw.

You then do the same thing with your other sounds on a second pass if you couldn’t track them all but track a kick again. You line up this kick wave exactly the same as your other one.

If you get any drive or latency it’s essentially just converts or your mpc natural jitter at that point.

This is the way a lot of people tracked without the outs to separate one set of tracks at a time.

Marco Polo even made a video on this same technique a few years ago.

https://www.youtube.com/watch?v=RVJJD0W ... ure=shared