Ok, i made the test for you. The timeline is set to seconds so i guess its not as bad here as with your computer setup, but it happens too.
Correct. I am not having the same problem (in fact, not sure if I have any problems). So, does that mean I am forbidden to try and solve the problem? A person with a problem and a person without same problem have more chance of fixing the problem than two guys with the same problem.
Support. Sorry, was talking to FDrummer and not yourself.
Live and Reason do EVERYTHING better than Cubase apparently. Infact, from the many references I see across this forum all the other DAWs are better than Cubase. 
Of course you can help! Where did i tell you the opposite? Damn internet…
I love Cubase, its the most complete DAW on the MIDI side and thats why i use it but i have this big issue with something as basic as recording notes from a controller. Totally paradoxical.
A lot of others don’t think so.
A midi controller? How does it record bare data other than notes? Pitchbend etc.
Checkable in the List editor. Does that come in late as well?
Yes.
Thanks. I’m still mulling that over.
So it records notes and control data late.
Is there another way to input notes and data for you? ie: Do you have an onboard soundcard to use temporarily to use to see if the data arrives or is “corrected” late.
I’m thinking in some cases it could be soundcard driver (but not discounting Cubase either) related although Cubase should correct any timing errors. If a different card produces other results then it puts us closer to the cause.
FWIW, did the test for comparison (no UAD in my system)
Result:
@32 samples buffer size:
Midi data varies from 144 to 192 samples earlier then audio.
@1024 samples buffer size:
Midi data varies from 1376 and 1968 samples earlier then audio.
Test was done at 48khz
I just got this from tech support.
"I am sorry but the latency correction works perfect and we do not have found any misbehaviour
until now. The recorded Midinotes are related to the latency time the ASIO driver reports and if the used VST Plugins have an additional latency this will be corrected too. But only if the used VST Plugin reports the correct latency.
Some older VST Plugins are not reporting the correct additional latency. When the latency report of an entire VST Plugin is not correct, then it is not only track- affected, the whole project could be affected, especially when this defective VST Plugin is inserted in the Master- Output.
And in fact we don’t have other customers reporting this issue and our QA doesn’t find any misbehaviour at all,
nor with UAD- Plugins used on the track or either with using any Steinberg VST Plugins.
So this issue has to related to the entire system, software or computer.
Best regards,
Technical Support Unit"
Hi guys,
thank you funkydrummer for copy-paste my email.
We don’t have found any misbehavior at this moment. The UAD- Cards and VST Plugins don’t show any issue related to the Midi- Recording.
Greetz
Marcus
Thanks Marcus for the input.
Just out of curiosity. On what kind of system did you test?
Clearly some systems are failing with latency correction, so for personal troubleshooting, it might be very helpful to know the specs (HW/Audio Interface/controller/OS) of a stable testing machine.
Mine tested on Intel i7 920 at 2.67gHz / Abit AI7 / Nvidia display / Emu 1212m / 12Gig ram.
Although many variables could cause this from bios settings through other installed apps and drivers to bad power supplys. Thanks to staff for the input. Very much needed sometimes.
Hi folks
Nice thread here!
Yes it is fundamental in Cubase:
It has to show us accurately what we have recorded.
Here in this thread we intend to play a key in a determinated point of time in the sequence.
And we expect that Cubase plays it back in exactly the same point.
Thus Cubase MUST be able to set the MIDI note at the same place like the audio recording of it.
That means, when playing back the MIDI track, the note should show where we have pressed the key,
but playing back, Cubase must automatically trigger it somewhat before, because of the
latency.
So there is a difference between what you see and hear and what Cubase does under the hood.
That what’s under the hood, is the task of the programmer.
And that what we want to see is what we are going to treat now.
So let’s bring some structure in our determinations!
What we must do is proving Steinberg, that there is a bug in Cubase.
I apreciate very much the test proposed before, to record a MIDI track sent to
an instrument > Group track > Audio track. (From funkydrummer.)
To examine if there is a bug in the program, there are too many things together
in the test project. Let’s go on in several steps looking only at one detail at the time.
I plan to divide the following steps:
-
MIDI recording without instrument
-
MIDI recording with an instrument
-
MIDI recording sending the instrument’s output to a group track
-
MIDI and audio recording like proposed before
In this way we could make some considerations or conclusions on every single level.
1. Step
-
Create a new empty project
-
Create a MIDI track
-
Do NOT connect it to an instrument!! (Don’t load any!)
-
Switch the click on
-
Play along some notes on your keyboard and record them
-
Look up in the Key Editor where they are placed.
Near the beat boundaries? Mine aren’t!
When I switched off the “All MIDI In” in Device setup/MIDI
and in the MIDI track I changed the Input directly from my “Fireface 1”
and then recorded another part… then the notes where veeery near
to the beat boundaries (quarter note). There, where I played them.
Yes! As it should be, showing only MY human feel and not the one
of Cubase.
Conclusion:
This is a BUG!!!
“All MIDI In” does not work well.
Workaround: Use a determinated input of your sound device.
Do not use “All MIDI In”.
If you play around with the Timestamp switches, you will notice,
that again your notes will be placed in wrong points.
Another BUG!!!
Workaround: Leave these switches = OFF
2. Step
-
In addition to the WORKING (!) MIDI track from Step 1 load now an instrument
-
As in the proposed test I loaded NEON with the “Blip” preset
-
Do still not open any Group track or Audio track
-
Record some MIDI notes like before
-
Look up in the Key Editor where they are placed.
Near the beats? Nooo!
Now they are distant on the left side (early).
Exactly like in every other test that has been made in this thread.
Conclusion:
This is another BUG!!!
Whenever you open a plugin, Cubase places your MIDI notes earlier.
Workaround: Hm, shall we get angry? This is an issue that exists as far as I remember
since Cubase SX1 or 2, at least when Steinberg introduced the Button
“Limit latency compensation” in the toolbar of the project window.
Yes, the workaround… weird…: In the info bar in every MIDI part increase the
start point by the mean offset of your MIDI notes.
Mine is 700 Samples/15 frames for a project with 120 bpm.
I think this could be different for everyone of us.
Depending probably on:
- Buffer size of your sound device
- Samplerate of the project
- Metronome etc
Don’t forget afterwards to snap the start handle of the part to the bar.
Then enjoy your professional program…
Well, this one for today.
Try it and let me know…
I will come back in the next days…
No problem here.
Create MIDI tracks, one with ‘All MIDI Inputs’ the other ‘direct MIDI source in’
Zero difference between both tracks.
Again no problem here.
In fact enabling or disabling doesn’t have any effect at all in my case (Windows MIDI).
Again no noticable problem here.
I used a 1/4 apreggiator as MIDI input device for this test (more steady then playing it by hand) and switched neon on and off while recording. The result: No difference in MIDI timing whether neon was on or off.
My system is pretty old and not rock solid with its MIDI timing, so when you tested this on a fairly new machine I’m afraid your system has a BUG 
I can confirm this is a problem for me too.
When I record anything in from my Edirol PCR-M50, it all shows up exactly 1/16th early on the created midi clip. It’s a doddle to fix, I just literally shift everything 1/16th forwards.
Snap is turned off, should be no auto correlation at all, but still goes in fractionally early.
There is an option to adjust for record delay or something, you can shift the record time by a number of samples, which is of no real use as I’ve no idea of the relationship between samples and beats.
Cubase 5 has it, Cubase 6 has it, but FL Studio has it.
I asked Edirol, and they blamed Cubase, I asked IL they blamed Edirol.
I think there’s something to be said for USB connections for midi controllers tbh, I think this is the problem.
But like I said, it’s literally 1/16th early, so fixing it is not hard at all, hence why I don’t mind it so much.
I tested it with a click track. Just hitting the C key in time at 120bpm, doesnt matter about accuracy, in 30 bars you are bound to hit or be extremely near the beat, but nope, everything was early.
I would suggest it’s a case of sound vs software. You hear the note in time, but from the time you put any pressure on the midi keyboard the signal was sent, it’s recorded. Difference should be fractional, but who knows, maybe that’s the problem, 1/64ths arent small enough for recording.
My thread
basically seems this same problem
I think the easiest way to sum it up is that you don’t hear back what you played, which indicates something is wrong. At first you doubt yourself but sometimes its so noticable theres no doubt about it.
We all know latency affects timing, but it affects the time between the key hit in the real world and the audio coming back out after being processed in the digital world. If it leads to notes being recorded early then something is wrong with something
Im clearly not the only one suffering this problem. And I have open mindedly tried every suggestion from the knowledge base and from users on here, and nothing has really cured it.
Can I ask everyone…
Are those of use experiencing this using Firewire or USB interfaces ? I personally didnt have the problem until I got a firewire interface. (but this was around the same time as my move to Cubase 6 also).
I am able to work around it, but for some, this problem must be more than just annoying and perplexing
It happens for me if I use a firewire interface or an internal sound blaster tbh.
Hi All
This topic has had me wondering so I tried an experiment of my own. I recorded myself tapping a single note to the click in both reason and cubase at the same time. I then exported the midi file from reason and imported it into cubase.
Next step was to select the reason midi file and lower the notes by using the arrow keys so as not to affect the positioning. I then could open both files in one editor and compare the results.
I tried this at various buffer settings.
After reading this thread the results I got were exactly as I expected. There is a difference in the way the 2 programs record midi.
Regardless of the buffer setting cubase records my midi hits as I press the key.
Reason records the midi as the note sounds so the higher the buffer setting the larger the gap between the two notes became.
Cubase consistently showed my hits on or as close to the click as I played but in reason they began to show up later and later with the buffer increase. ( I was trying to play on the click in the real world and not alter my playing due to latency)
I personally prefer the method Cubase uses to record midi. It allows me to still record in on time when my buffer setting are really high, by turning down the track volume and monitoring the audio output of my keyboard. My keyboard has on board sounds and I monitor through cuemix on my motu so I always have near zero latency on everything, even at a 2048 buffer setting. Great when you need to add stuff to an almost finished mix that couldn’t play at 128samples.
However I can see the shortcomings for those without the kind of set up I have. Perhaps a prefence to record midi as it was monitored would be welcome by some. At present it records as it was played in the “real world”.
However if a permanent change was made and Cubase behaved like reason I would be a very saaaad Panda.
Options are best in this sort of situation.
Anyways, thats my 2 cents worth and considering I live in South Africa at the current exchange rate its about 0.2 cents everywhere else 
Peace all…over and out.
Forget about the kind of interface. It happened to me always in several versions with pci&firewire interfaces.
Related to USB ports in controllers? Might be, but i think over the course of these 6 or 7 years i tried everything so i doubt io would have forgotten if soemthing really solved the problem.
Ali Baba…i´ll try your tests and post the result back.
Really depressing that the official answer is: its related to your computer. I wish i lived in Germany and could bring my computer to the Stein´s support office just to show them. Would they look at this problem then?
If Steinberg cannot make a product which records MIDI on a way thats acceptable on every decent computer system then they should put it clearly in the box: use Cubase with THIS processor, THIS controllers, etc.
the point here is there´s many people getting recorded MIDI earier than they played (latency doesnt matter here). I wish it worked here as you say its working there
Ah! I use PCI interface. Always have. I’m about due to upgrade so I’ll keep a wary eye on this thread. Could be handy if there is an unspotted bug with externally connected cards.
I have noticed that there has been a lot of corrections especialy for usb / midi soundcards over the years. In fact I think the “Constrain delay” and timestamp were usb related which is why I stuck by PCI cards.