I’ve been using Cubase since the Atari and the Steinberg SMP-24 with old school 24 track tape back in the Naughties. For example, we recorded this version of a well known funk classic at Trident Studios in London back in the day with old school 24 track synced up to Cubase via a Steinberg SMP-24 via a SMPTE stripe. I did have to debug a problem where the SMP-24 had a stored program where it changed tempo by a couple of BPM after XXX number of bars but once fixed it was ROCK SOLID. We laid down drums, bass, clav and guitar to the click and some MIDI triggers. My clav wobbles a tiny bit on the intro (that’s my fault) but apart from that, we’re “tight as a gnat’s chuff” and in the pocket.
We laid down everything else, including my congas on subsequent passes and there was NEVER any concern about timing. In fact we just didn’t think about it. We were focussed on getting our playing in the pocket and the gear was rock solid.
I hear people in these forums saying that nobody can hear a 10ms delay and reciting crap maths at us.
** 10ms is a gulf between “The Pocket” and flabby shite**
5ms doesn’t cut it either. VST pianos are only useful to me as instruments with under 2ms latency. The problem I’ve got is that, even though I’m older and not as fast as I was, I’m noticing that layered DAW multi track recordings have VARIABLE record latency on EACH track in DIFFERENT places and that it wobbles both on each playback and record so setting a timing offset on each track just doesn’t cut it.
Wow and flutter on old tape machines would be across all the tracks so if I was laying down (for example) the congas on top of the backing track, and the tape drifted, I could feel it and compensate. That process was entirely subconcious. I know I’m getting older, but these days, every time I lay down some drums on an existing Cubase track, I find myself thinking: “I know I was in the pocket when I played that” and worse, every time I play it back my pocket seems to be off in a different place. It’s only when I do the mixdown that I hear any consistency of “outness”. I’ve spent hours fiddling with timing offsets but the problem is variable. It’s effectively wow and flutter on each separate track. I’ve tried different ADC/DAC units to no avail. I’ve noticed it on multiple different hardware setups on both Mac (Intel) and Windows.
I have this nasty feeling that it’s way down in the hardware layer. The whole concept of interrupts is stuttery by definition…
I wonder how much that whole genre of out-of-pocket-in-pocket grooves has come from all that…?
I would love to get this resolved but I have a nasty feeling it’s only going to happen when processor and bus speeds get to microsecond timings for audio latency…?
I’d be really interested to hear from anybody with constructive, practical suggestions. I’m not much up for being gaslit by theoreticians !
Current rig:
Cubase 14 Pro
Windows 10
AMD Ryzen 7 3700X
ASUS ROG STRIX B550-F Mainboard
Motu 828 USB
The latency you mention is not to do with Cubase unless you have plugin inserts on the tracks or the main bus. Then the extra latency is due to the plugins. When recording midi make sure there are no plugins as mentioned above. Disabling them does not work. It’s best to remove.
The rest is down to your audio interface and computer. I can feel latency but it feels tight to me at 6ms playing drums. With keys I can go a bit higher. I suspect you have something in the path causing the latency as the audio interface latency won’t be changing. I use RME which are good down to about 3ms if the computer is powerful enough.
Many thanks for your commentary.
I’ll try with all plugins removed. If that is the case, it’s going to make the production workflow very difficult…
I can feel latency at 2ms for both drums and keys.
Cubase is reporting 3ms latency with the Motu I notice that display flickers which means that the latency calculation is not static and depends what’s going on in the system. Which makes sense. That’s why it also makes sense that the latency is variable.
To be clear: The latency values of the ASIO device are reported by the ASIO device to Cubase. Cubase does not calculate anything. It merely displays the value that the device has transmitted. If the device transmits a wrong value than Cubase will display a wrong value.
There should not be any flickering as this type of latency is static. It is my understanding that manufacturers of audio interfaces have a table inside the driver for the measured latencies. I really don’t think any device “calculates” latencies in real time.
You can check out other DAWs and see if any of them yields better results in terms of timing.
If not then music production through a modern DAW might just not be your way to go and you might have to find another way of creating, recording, producing music.
Sometimes the crap math is quite useful, though.
The following is not meant to diminish your point. I merely write for others to bring some perspective what we are talking about when mentioning milliseconds of latency.
In a studio the sound needs 3ms to travel ~1.03 metres. Ie. if a e-guitar player listens to their guitar playing through a speaker and they change the distance from 1m to 2m they have doubled the latency from 3ms to 6ms.
Just an info for perspective.
A drum player would have a steady distance to the drums and therefore a steady latency. Also there, however, the kick drum has more than twice the latency of the snare drum. And we are talking about a difference between those two drums of 1ms or 2ms.
@Johnny_Moneto provided helpful infos in a respectful, condensed and meaningful way to put things into perspective. If not for you, than for others since this is a public forum.
It has nothing to do with gaslighting, crap or irrelevant infos.
Nobody elaborated on musicpsychological or neurobiological concepts/findings to question your personal definition of timing within said range or your individual perception of groove.
That would be fair enough if I had not SPECIFICALLY asked to be spared the crap maths and theory in my OP. I remember when Digital Audio first appeared and loads of HiFi purists were complaining that they could hear noise problems. The theorists told them it wasn’t possible because (“insert bad science/theory here”). Then the folks at Prism Audio and some other high end audio folks ACTUALLY looked into it and discovered noise artifacts caused by heterodyne demodulation due to clock jitter. Suddenly the noise wasn’t down to some nutter with “musicpsychological or neurobiological concepts”. Somebody had finally figured out what the people with ears had been hearing all along. Suddenly they started getting interested in clock accuracy and all of a sudden the sample rate went up as well. So, no, when somebody starts telling me what I can or cannot hear based on some crap maths then I am going to call them on it. Please note that I did thank Mr Moneto for his very helpful practical suggestions.
I think Cubase is a fantastic piece of Software. I’m just trying to find a way to use it as effectively as an old school 24 track !
While it’s irrelevant to you, it’s relevant to 99% of the rest of the world who may try to help solve your problem. Since auditory cortex processing alone can take 4-5ms even in highly adapted cases of use-dependent plasticity, your ability to discern 2ms latency in keys/drums makes it impossible for most people to know if what they suggest will even help in the first place since they won’t be able to hear it. So “the crap maths” have obviated an entire population’s ability to even contribute. That kind of shifts it back into “relevance.”
Folks have made other helpful suggestions than the crap maths which has a subtle inference of “What you’re describing is impossible so you must be a nutter”. I refer the honourable gentleman to my OP and tales of old school 24 track analogue tape machines at Trident Studios in London. I really do WANT to use Cubase. This all may be more relevant than you realise. I might have to start quoting Richard Feynman in a minute !
Your best bet is to get a really low latency interface if it impacts you that much. Also drop the plugins if they cause latency. There were restrictions how you could work with tape and there are restrictions with digital. If you want to use as a 24 track then there is no problem with audio as long as you use direct monitoring and only use the effects on the interface. That’s how I do vocals. With vsti’s you really need a low latency interface and a fast computer as mentioned before. The restriction in this case is no plugins that increase latency while playing.
I must admit I struggle that you can hear 2ms latency. If you create a single vsti track with no plugins on the track or the master can you still hear it? Also not all vsti’s are equal in latency. Not all midi devices are equal either. The Roland drum modules have lower latency than say an Alesis module. I’m talking the time it takes from a hit to generating the midi. So depending on what you use to trigger drums can make a big difference. My guess is keyboards are the same. Now this problem would have existed back in the Atari days. Also back then there was problems with the time it took to generate the sound from midi.
It would also be interesting to know whether you record natural percussion through microphones or whether you use electronic percussion sound modules. Especially in case you are using MIDI that can make a huge difference for a timing sensitive person.
But he is (also) claiming that the problem is “variable latency on each track, and that the tracks each “wobble on playback”.
This should be easy to test: record a click track in Cubase (metronome e.g.), verify that is precisely in time. Now record that track back into Cubase with a loopback cable from output to input . Any variable latency should be immediately visible.
It would also be interesting to know whether you record natural percussion through microphones or whether you use electronic percussion sound modules. Especially in case you are using MIDI that can make a huge difference for a timing sensitive person.
I record live percussion. I don’t use external MIDI modules for percussion.
I record keyboards via MIDI and thru to my Nord Stage 2 which is rock solid real time.
However, when I finally record the output of the Nord back into Cubase, I notice some very subtle timing differences on the resultant playback. Most of the time it’s not quite enough for me to do anything about it… VST instrument playback seems really consistent. It just seems to be the recorded, tracked audio that seems to have subtle, variable shifts in timing in the context of everything else.
Some studio cats swear by having high end clock generators driving their audio interface.
Interesting. How does Cubase fit in with that ? Can I get Cubase to take clock from the DAC ?
Or maybe even a separate super stable external clock ? My Motu 828 has a word clock input.
Also not all vsti’s are equal in latency. Not all midi devices are equal either. The Roland drum modules have lower latency than say an Alesis module. I’m talking the time it takes from a hit to generating the midi. So depending on what you use to trigger drums can make a big difference. My guess is keyboards are the same. Now this problem would have existed back in the Atari days. Also back then there was problems with the time it took to generate the sound from midi.
Yes, I can remember noticing timing differences with my Roland JV80 compared with Roland D10, D50 etc over MIDI. For the most part the MIDI delay offset would take care of it because it was reasonably consistent. That has all improved massively. My Nord Stage 2 goes like the clappers over MIDI. This stretchy audio thing is a different wrinkle…
