I recorded audio simultaneous with the MIDI from my Motif into Cubase. My audio latency is theoretically 5.8 msec, (though according to Cubase, the input latency is 6.531 msec, ). As expected the audio straight from my Motif appeared consistently after the MIDI: 2 msec, 5 msec, 5 msec, and 4 msec later. Assuming consistent input audio latency of around 6 seconds then, it seems the MIDI was recorded (inconsistently) between 1 to 4 msec after the key was hit.
Is this inconsistency, “MIDI jitter”?
And why is the calculated latency of 5.8 msec (256 samples, 44.1K) different than Cubase’s reported 6.531 msec? And why does Cubase report different latencies for input (6.531 msec) vs output (6.485 msec)?
Thank you Conman, I think it is … I dug it out to reread before posting!
But the reason I’m asking is that I thought it could be any one of a number of other things I can’t even think of that might be causing that behaviour.
BTW - do you have any insight into the “Input Latency” and “Output Latency” reported by Cubase in the Device setup window not being the equal to each other … as well as being different than the “calculated latency” (buffer size/samples per second)?
Input & output latency is a round trip total through the computer. Components probably give the difference but if you download one of the free latency checkers you would probably find that the two figures add up to your true system latency.
A latency checker should tell you what is causing any data blocking.
Jitter is mostly a hardware issue. Software should generally correct jitter and not cause it. A lot of users get over-excited by it and get needlessly angry at software they think is causing it. All depends on little clock circuits disagreeing with each other.
Have you checked “Use system timestamp” in the Devices setup? Might improve that timing difference of the midi / audio.
I see from your spec (good man) that, unfortunately your system may be somewhat underpowered for Cubase 6. Not enough ram for string libraries, XP is not supported by Steinberg (but Cubase should work fine under it) and the computer is old which means that any data from the ram on that model goes through the “Northbridge” which translates data from the ram to the CPU & V.Versa. which is typically where you get holdups.
Newer i3/5/7 systems use direct communication between ram & CPU so things are much improved.
Cubase is now optimised for i3/5/7 systems and Windows 7.
Treat yourself.
I have 4 hardware synths and each have their own latency amount. Some jitter more than others too. It’s just the way hardware is. It takes time for the MIDI to go from Cubase to the synth, then the audio from the synth to Cubase. That’s what I believe anyway. I just set the MIDI track’s delay to -4 ms or what ever the amount of delay you get. Not much else you can do.
Hey ilmolto - thanks for that. It’s good to hear that the multi-millisecond delay is not unusual, or at least it’s seen in one other system.
I’ll look in the preferences for a global MIDI track delay - I’ve only got one synth, so it’d be nice if I could get by without having to set the delay for each and every track I might add during a project.
Conman - I’ll have to re-read (again!) that article you kindly referenced, I came away with the impression that a big part of the jitter was the way Cubase bins the incoming MIDI data and then spits it back out … i.e., not just a hardware issue.
Jitter is a rarity in software. Cubase does seem to have some timing issues for some but it hasn’t been firmly established what the causes are. Could be some systems are more prone to it than others but Cubase is a large program with many settings to compensate and the shedload of features are probably a headache to trawl through to find out if any directly cause, or fail to compensate for, system timing discrepancies.
If you haven’t tried it there is also in Devices setup a “Priority” setting which alters the priority of audio and midi data (used to be 4 stages but now only two). You could try setting that to midi. I think it defaults to audio priority. Whether they would arrive at the same time though is anyone’s guess.
Imolto’s idea is good as well.
You could also try using the DAW Reaper (free unfettered trial) to see if the same timing applies there.
But my final judgement, for the way I work anyhow, is that if you can’t hear it don’t fret too much. I know that doesn’t apply everywhere for everyone but if you need that accuracy it’s really better to shop around to find something that does the job that way.
I’ve yet to go into a studio where a bit of headscratching doesn’t happen at least once a day with any recording medium or DAW. At least things don’t go bang so much these days.
The article pointed to is from 2007 so most of the points may have been superseded but I think the background relevance is useful to give a (partial) picture of what’s happening in a system.
Never record midi unless you have to…or as some kind of hail Mary safety. It’s not accurate…on any platform…period. Compare 980 ppqn to 96,000 samples per second…or even 44,100. “jitter” is a tiny factor compared to the lack of resolution. It’s rounding whatever you play to the nearest ppqn.
I’m just saying…I sequenced, sync’d to tape for years…I could not stop fast enough once digital audio made track count expand to have enough economically to record what I played.
Recently tested as proof on another forum…recorded the live audio out of a VI, as well as midi similtaneously. Then bounced the midi in real time to audio. Then bounced it offline to audio. Three different results. I’ll take the first, because it’s what I played.
Cheers, that’s real life for you.
Interesting in the light of various comments that “Ableton does it…” comments that are often seen here.
And the best studios do run smoothly for more like a couple of days or three if you’re lucky but even live lots of things go wrong nearly every gig. Strings go, leads, valves, pedal springs and non-electrics like drum heads go. So, if everythig’s running kind or smooth don’t worry about the odd minor glitch too much but at the same time nothng can be ignored so it’s ok to come to a forum now and again to investigate and even have the odd steam-off moan.
Hell I’m just wrestling with a soundcard that lost outputs and needs it’s teeth cleaned at the minute. I had to look at a few forums to make sure I didn’t need to reinstall Cubase.
Yo quote an old teacher. It’s not what you know it’s knowing where to look for what you don’t know that solves problems.
). As expected the audio straight from my Motif appeared consistently after the MIDI: 2 msec, 5 msec, 5 msec, and 4 msec later. Assuming consistent input audio latency of around 6 seconds then, it seems the MIDI was recorded (inconsistently) between 1 to 4 msec after the key was hit.