Digitakt Sequencer has the jitters

Elektron Gear Digitakt
#1

Hey all, lately I’ve been recording a ton of tracks from my Digitakt, Syntakt and Analog Four mk2 into my DAW. I’ve found that there is a slight drift or jitter on the audio files that I record from Digitakt. This problem has been twisting my melon for the past few months and after an obsessive amount of testing and head scratching, I discovered what is really going on. I’d like to share this with y’all and see if anyone out there has experienced something similar - or knows a solution, because I’m pretty share there isn’t a perfect one. Here goes:

First off, I’m recording all tracks (one at a time) from Digitakt via analog output (24bit/48k) - not digital. (an explanation of why I don’t use overbridge to record is below)

Secondly, I’m not using MIDI clock to sync at all. I only send a Transport start message via USB to Digitakt to ensure that I’m recording only the internal sequencer clock from the Digitakt itself. Obviously elektron sequencers sound amazing on their own and I want the DT’s internal sequencer locked to the correct tempo. MIDI clock receive is set to off.

So after I record a section of tracks analog with no external MIDI clock sync interference, I get the downbeats of the tracks perfectly lined up on Ableton’s grid and when I scan later in the audio file, very often there is a small but noticeable drift in both the +/- direction throughout the recording. This becomes a problem when you have 8 DT tracks playing at once, because all those little timing inconsistencies add up and smear the groove.

With me so far? Ok, so regarding Overbridge:
Overbridge is actually the solution I’ve found for this jitter issue, but its not a perfect one. When I record multiple tracks simultaneously from Digtakt via Overbridge, ALL of those tracks have the same exact jitter/audio drift. So at least the jitter is consistent within that section and the groove is maintained. If I do another recording of the same tracks, they will have a slightly different jitter than the previous take, because the the Digitakt sequencer plays it microscopically different each take. But those tracks in the new take will all be in sync with eachother.

However, despite popular opinion, I haven’t been using overbridge to record, because the recordings sound way better/juicier/fatter/warmer when I go directly out of the analog outputs of DT into my audio Interface/preamp. Recording multiple tracks simultaneously via overbridge is super convenient, but it loses its sonic mojo compared to how it sounds when I’m playing directly out of the DT. Also, when you record multiple tracks at once via overbridge, you can’t print individual track Send FX on or with the individual tracks - as far as I know?

Bottom line: Digitakt (and Syntakt) sequencers have a slight, almost imperceptible inherent drift/jitter. This is probably intentional by design and is part of elektron sequencers’ secret sauce. Recording tracks one at a time out of Digitakt’s analog outputs is tedious, but it sounds way better to me than recording digital via USB through overbridge. The trade off is that I’m having to go in and warp most of my DT rhythm tracks in ableton to correct the audio drift and try to glue the groove back together - which is insane.

I couldn’t find anything anywhere talking about this specific issue… I’m sure someone out there has noticed their groove is a little wonky after they record it… and now we know why. I doubt Elektron would ever admit that their sequencers have inherent jitter… haha. BUT they could implement a way to record multiple tracks via overbridge and print the FX from the individual tracks separately or on the track itself - all in one take - then that would be the best solution for preserving your groove AND also being able to edit the recording with FX.

1 Like
#2

There has been another thread going about jitter here: https://www.elektronauts.com/t/psytrance-discouraged

Everybody seems to different opinions about how to deal with it, and how important it is. Jitter kind of drove me crazy for a bit. I tried a bunch of things, and in the end, what I always do now is resample the kick and snare, but I allow jitter for the rest of the sounds.

I do that, by just throwing the kick and snare into samplers on their own tracks, and and quantizing the midi to 1/256. To me, that amount of quantization makes no difference to the grove, but fixes the jitter.

Also, to get the midi from the elektron box, you can record midi from it, or do what I described on that thread, and just extract the midi from the audio file.

For the rest of the sounds, I think it is ok to leave some jitter, unless maybe you are using bass sounds in your digitakt. I do all my bass in the box, so that isn’t an issue for me. I don’t resample other tracks besides kick and snare, because it is too much of a hassle, since most other sound for me have a lot of parameter locks.

I think it is important to fix jitter for kicks and snares, because it can cause a lot of problems if you don’t. And, I like to have midi tracks that are exactly aligned to the kick and snare for sending to sidechains. If you leave jitter, your midi and audio tracks won’t be aligned.

I would not use audio warping to fix jitter, because it will bring down the audio quality and smear transients on drums.

#3

thanks for all the great intel! yeah I’m hard-quantizing kicks and snares to the grid like a beast… even hit hats that I know are supposed to be exactly on the grid upbeat. Let all the in between notes do the grooving.

That’s interesting what you said about recording the midi from the elektron boxes. Do you use the midi to then trigger the parts from the original instrument or resample?

I used to be a mega psytrance fan back in the day - like early 2000’s… when Hallucinogen, GMS, Astral Projection, Yahel we’re the guys… Had like at least 12 Distance to Goa compilation CDs lol. Send me some music links if you want… your stuff or artists you would recommend.

Cheers.
-Taylor

#4

Hi Taylor, I was using midi from my Elektron box (mainly machinedrum) for a long time. But recently, I found that it worked better for me to just extract the midi from the audio. I do that, because the jitter can cause the midi to vary from the recorded audio, so I think I get closer timing by extracting. Many Daws extract it easily. I use Bitwig, so I extract midi from the Audio with this basic patch I made in Note Grid (below). I quantize to 1/256, because that fixes the jitter, but doesn’t mess up the timing if I have a lot of swing on, or something like that, and keeps it sounding like it did originally. Again, I like this method also, because I can use the quantized midi for sidechain triggers, which wouldn’t really work if I didn’t fix the jitter issues. I don’t make psytrance, so I’m sure the techniques are different for it. I produce Glitch hop/ bass music/ breaks. Last thing, before doing any of this, I select all tracks, and try to line them up as close to the grid as possible, to make up for any latency.
Cheers
-Andrei

anrei_note_extractor
anrei_note_extractor817×452 26.4 KB

#5

Each to their own and all that but this all sounds like so much work

Have you tried simply clocking your Digitakt to the rest of your gear and recording it like a normal person?

8 Likes
#6

I love the idea of being a normal person, but I fear for the risk of living life with jitter issues.

3 Likes
#7

jitter

(Sorry I don’t have anything constructive to add :laughing:)

1 Like
#8

That’s good reminder I need to go re-caffeinate, and continue to zoom in very closely on waveforms.

3 Likes
#9

What are we talking about in milliseconds? Do you have a screenshot of the drift, say 1 minute into the recording, with waveforms that should be perfectly aligned? Would be interesting to see.

#10

1.4
1.41410×1464 168 KB
2.4
2.41750×1596 270 KB
6.4
6.41560×1464 217 KB
12.4
12.41430×1594 218 KB

Here are screenshots of a 16 bar kick and clap track… they are aligned on the grid perfectly at 1.1. There tracks are not warped.

Over 16 bars each kick and clap which should be dead-on the grid are slightly moving ahead and behind throughout the track.

2 Likes
#11

And for someone not familiar with the Ableton interface. Are we looking at milliseconds here?

#12

Curious if things would be any better if you recorded from the DT in class-compliant USB mode rather than over analog.

By using the DT as Live’s audio interface, you’d force DT and Live to see the same timebase (samples per second) as the basis of their clocks. So, while the Elektron sequencer might not be mathematically perfect, at least there would be no drift.

I used to use this same technique with my MPC 1000 and clocking my interface via S/PDIF, and it allowed me to get “perfect” sync of recorded tracks against the grid without actually having to sync anything with MIDI as you describe.

#13

Yes. Milliseconds.

Seems insignificant, but when you have 8+ tracks going simultaneously, all with their own slight drift, it’s gets squirrely

#14

All right! Each to their own I guess, but I’m actually quite happy to see that this is the case, that there is slight drift that is.

Many of us are trying to find ways to give our ‘perfect’ gear with sequencers more of a human feel. So to see that it is built in, intentially or not, is great news to my ears!

So do I understand correctly that the philiosophy of psytrance is to have it surgically precise on the grid? Sorry, but can not understand the appeal of it in that case unfortunately.

1 Like
#15

Definitely have noticed this as well. Couple of things that have worked for me:

  1. adjusting your midi latency, for just the digitakt, in abletons preferences
  2. setting the first hit (first gold transient) to be on grid in abletons clip editing post recording, thereby quantizing the sounds following to be as they were heard when recording

No matter what, if you are using groove and swing in ableton it will change the groove and requantize it accordingly by transient.
In the end, I personally love the “jitter”. it definitely gives some character and swing to my rhythm.

1 Like
#16

There are some great ideas in this thread. I have been exploring the use of “slice to midi” for kick and bass lines captured by overbridge - and using this to quantise them

An Alternative Idea

what about keeping the jitter (since the majority of work done is in hardware drum machines) and instead extracting the “groove” from overbridge captured phrases to apply to DAW MIDI clips? These adjusted midi clips will trigger software synths and other things in line with the slightly jittered overbridge samples

Theoretically transients will line up and so will MIDI side chain triggers

I know Ableton had specific warp algorithms for drums which preserve the transients as well for normal audio quantise

#17

I have experimented with doing things that way, but in honesty, it made everything more difficult for me. Because, it was just a big hassle to align everything else to the drifted midi. Mainly, because the drift is so inconsistent. I tried starting with aligning all my bass hits, but it meant every one would have to be shifted earlier or later. So, it just became too tedious.

So, that’s why, I still think it is best to resample the main drums (kick/snare) and quantize the midi (to 1/256). Again, I found that small amount of quantization to fix the jitter (usually), but not mess up the groove/swing. The only problem for me with this method is if there is variation (p-locks) withe the kick/ snare. For most of my tracks, all the p-locks in the drums are on other sounds. So, that’s why I only quantize the kick and snare. The drifting hats/perc on top of the kick and snare could possible interfere with the transients a bit. But, I generally have some sidechain on my perc/hats to keep the kick transients cleaner. This isn’t really necessary in a lot of styles, but I think if you want punchy beats, and it will be mastered loud, that it can be important. Wither way, I can live with some jitter on my hats/perc, but it is just too much of a problem on the kick/snare.

Like I mentioned earlier, sidechaining is another reason why I prefer to have quantized midi triggering the kick/snare, rather than shifting the audio. I always use midi for sidechain triggers, because it gives a lot more control. I do it in Bitwig, but sidechain vst’s like “Duck” or “Gatekeeper” accept midi, and can be used in any DAW. For me the advantages are being able to use an envelope to control the sidechain, and also being able to shift the midi that is triggering it, to be a little earlier than the actual kick. That way, the bass is already ducked down a bit before the kick hits.

Anyway, I’m rambling on, but it’s because I am an obsessive nut when it comes to jitter. It kind of drove me nuts for a bit. So now, after recording parts from my elektron boxes (machinedrum/ octatrack/ digitone), the first thing I always do is try to align the time to be as close to the grid as possible, then resample the kick/ snare to new midi tracks, and quantize them to 1/256. That fixes the main drums, and gives me midi to use for sidechaining that is lined up perfectly with the audio.

1 Like
#18

I think you and I are very much kindred spirits but you have a lot more hands-on experience I think to solve this

I also use bitwig, duck and some of the tools I mention. I like the psytrance style and certain elements definitely have to be locked to the grid for this to work

Currently playing with Octatrack glitch sounds for this genre

Philosophically once you get down to sub-millisecond jitter things get into the quantum non-locality space anyway.

Examples:

  • Where does a sample really begin? Is it the first positive deflection after silence - some of which have such low amplitude to measure? Even beat detection algorithms struggle with this
  • if you use a “modulated delay” to avoid phasing you are going off grid with the delay hits
  • with high frequency sounds, (eg hats) “smeared” transients would be perceived more as a change in timbre than actual timing irregularities- also Eq and various common effects would have smearing effects - don’t we want the timbre to sound varied which is why we humanise velocities and decays?

Does Ableton “quantise with preserved transients” keep them intact? This would be a hell of a lot easier to line up percussion to the grid

#19

I hear what you are saying, and I don’t know how bad jitter is on digitakt. I think the jitter I get is worse than sub-millisecond though. Like I mentioned before, my Elektron boxes are Machinedrum, Octatrack, and Digitone. Out of those, Machinedrum has the worst jitter. But, I love using it.
As for when samples begin, and phase coming from modulated delays. Even though I obsess about jitter, I’m not too analytical about phase. My issue with jitter is more just that it is so inconsistent. I love using modulated delays. I love using a lot of fast retriggers too, that shift sounds around. And, I do granular resampling all over the place. So, for me, phase isn’t so much my concern.

The jitter I get seems to be a bigger timespan than would just shift the phase.
My main concern is that I don’t want basses that are supposed to be aligned with a kick and a sidechain envelope to be constantly drifting early or late. If there is a drop, and the kick hits late, the bass will hit before the kick, messing up the transient, and of course, any sidechains will be off. To me, if there are sounds interfering with a clean kick transient, especially on important drops (like every 4/8 bars), it just messes up the impact, especially once a track is heavily limited in mastering. An trying to shift all the basses to a constantly drifting time seems more difficult than aligning the kick/snare.

I actually recently bought an E-RM Multiclock. It is an improvement, but unfortunately, I’m still getting jitter on my Machinedrum. I’m just recording parts from my octatrack right now, so I’m hoping it will show more improvement.

I’m very curious to see if midi 2.0 is going to help solve these issues.

1 Like
#20

Your comments make sense. Hmmmm Is the jitter in your setup better if you record directly to the Octatrack? This bypasses the usb pipeline with audio interfaces. Is the jitter inherent to the machine drum?