[possible BUG]: Groove Agent One drifts in timing

@ Conman.

I don’t think it has anything to do with quantize, snap or anything related to that.

Here’s what I do:

  1. empty song

  2. create instrument with Groove Agent One and some preset. Create any pattern of sounds - doesn’t have to be a beat - just record some notes.

  3. I right click the track and then duplicate the entire track, using the “duplicat tracks” command. So the instrument and midi and preset chosen is an exact copy of the first track.

What happens is one of two thing, and it seems random:

  1. the track is duplicated and plays as it should: an exact copy so all is heard is a level raise
  2. the two tracks plays “out a sync”. It’s not noticeable as a delay but more as the typical “smear” when you for example do parallel processing with a plugin that doesn’t work with latency compensation. Sort of comb filtering. It’s quite nasty.

Exactly the same here, but mine is either perfect or has a noticable delay.

Sometimes it will play perfectly for a minute or so, then the timing error starts, then it might stop and correct itself or just continue. A bit like a dodgy random slapback effect.

Perhaps our different experiences are sample rate related. Tried testing @ 44.1?

I’m sure you are right. I deleted the original, started a new project entering straight 8ths using the line tool. Duplicated the track, same result. Horrible! Completely random too.

Guess what - I just tested @ 96 kHz and here it reacts just like Sturgeon descripes. There’s definately something samplerate related going on here!

Time to call the bug?


Yes ! Please try it @96kHz using the Honey Bee kit (short tight closed hat makes it more apparent),

If you don’t hear it right away, keep listening, sometimes it can take a minute or more. There is something strange going on !

I can confirm the behaviour but sadly maybe not the bug (but it could still be a bug). Here’s why.
After I compared two drum tracks I created a piano track, duplicated it and then put one track reverse phase. Result = silence, as you would expect. But being a little curious and remembering something forgotten in the mists I moved the stereo panners on each piano track and then I could hear the pianos.

My conclusion is that if all the drum instruments are not panned dead centre then you’re going to defeat any phase inversion test for silence.

But it’s only an observation of mine and just a part of the picture for you to take account of so don’t take it as done and dusted by me because I have my doubts on this too.
Could still be that some drum instruments are out of time for some reason. Maybe to do with the sample envelope start times etc. who knows?
So far I have not spotted a kit that is so out of time with other instruments that it hurts from a songwriting POV. I suppose it could affect very precise users.

@ Conman: again: the entire track is duplicated - thus a 1:1 copy of the original. There is absolutely no operator error possible here. All we do is right click the first track and chose “duplicate track”. There’s no panning, no volume, no inserts, no nothing is changed. There’s clearly something going on that shouldn’t be.

IT has tricked me for some time when layering kick drums, where only the smallest amount of drift can cause phasing and play nasty games with the lowend. Even a matter of a fes samples can completely mess it up - and when it’s random it’s a total mess.

You wouldn’t notice it under “normal” conditions when just playing one kit. It’s when you start layering and duplicating the “fun” begins.

I think its worse than that. Try this;

Create a new project @96kHz, Set locators to cycle round 4 bars.

Create an instrument track with Groove Agent on it, select a kit (Honey Bee is a good example).

Double click between locators to create a blank part.

Open part, turn snap on, and quantise value to 8ths.

Use line tool to fill part with closed High Hat 8ths.

Play, and listen for a long time.

No human note entry needed, or track duplication. Its ok for a while, then starts flamming and stuttering at random!

I did duplicate an entire track, I mean what else could I do? But I did see what you are on about. I just threw in an observation of my own. Not trying to disprove it but trying to see what is causing it in the case of C6. And I never assumed operator error, just investigating the possibility of maybe a sample inaccuracy in GA.
If you look at the note info for the same (ie: duplicated track/s) note in each track does it have a different start point? Sorry if I’ve missed that in the thread somewhere and you have done it already.

To recap on the stereo observation could maybe assume that, as in the piano sample, the fact that the sampled (pre-set, rather than your own samples) instruments, especially toms typically are allocated different pan positions which might skew an accurate playback. Of course, I’d expect that kicks and snare would be centred normally.
Though there does appear to be something funny going on as I’d expect, as on drum machines, that the layering of bass and snare particularly should not lead to a comb effect.

Conman: no worries, I just tried to make clear that human error was out of the eqation:)

And to make it even more clear I just did another test:

  1. blank GA1 with my own mono sample loaded as the only thing.

Bang: same result. Something’s rotten in the state of GA1. :mrgreen:

Did you try my test above Lasso?

Either way, you are right;

“Something’s rotten in the state of GA1”

Yes, I did, but it didn’t behave as bad as you describe - it stayed tight enough for me to not notice anything after 10 minuttes. However, as soon as the track is duplicated the timing issues become quite apparent.

Halion 1 drums work ok in this respect. The only thing I can think of to look at is whether either the samples themselves or the FX / filters applied within GA itself may give these effects by continually offsetting the samples so that no two ever sound quite the same.
In case you’re wondering why I’m apparently in denial it’s only so the guys looking at this have more details to hand to pinpoint a possible solution.

I suppose my next stage will be to strip out all the sample / pad settings inside GA1 and see if anything tightens up after that or not. May take a day or so.

I testes it with a blank GA1 instance with only my own mono sample loaded. Same result.

Hm. Looking more bug-like. Thanks.

Bump. Is Steinberg having a go at this? :mrgreen:

It looks like this is the only answer we are going to get.

Any chance you could try the test @96kHz using Honey Bee kit, and for a long time/several long tests Chris? It appears to be completely random and intermittent, so there is a chance that a single quick test would miss it

Its not just my ears, I’ve tried it on other folk and they agree that there is something very strange going on. Halion One/SE kits don’t behave like this, nor does anything else!

Sorry, byt this one concerns me a lot.

Could Steinberg please give a hint to wheter they’re checking this out or not?


Today I was working with a track that has several drums sounds spread across two instances of Groove Agent One.

Ahhh…how could I forget that Groove Agent drifts…thus making it impossible to use in conjunction with other sample players. Or more instances of itself?

It’s such a shame. Is Steinberg having a look at this?