MIDI notes recorded early in Cubase

Any light that can be shed on this long standing problem would be most gratefully appreciated. It may well not be a UAD-related problem, but UAD users are particularly likely to encounter it due to the latency involved in the data transfer at hi plugin counts.

I'm running a Windows XP Pro PC w/ MAudio 1010LT PCI sound interface, with MAudio Oxygen 49 keyboard (USB) UAD 2 Quad (6.1) plus a faithful old UAD 1

When recording MIDI in cubase, and I have a large buffer size set on my sound card (eg 1024 samples) there is noticable time delay between hitting the MIDI key and hearing the sound.
This is what I would expect . My problem is that if I compensate by playing early, the notes are recorded early even though they sounded in time when played live.
The problem becomes amplified as the buffer size increases or I add more UAD plugins. This is obviously because the data is having to do more travelling across the PCI bus through all the UAD plugins.
However, the host does not compensating correctly for the total latency. As I increase the buffer size, or add more UAD plugs, I have to anticipate the note even more, and so the notes are recorded even earlier in the sequencer. This is a right royal pain in the ass !

I've tried everything I can think of in cubase (including using system time stamp, and trying emulated versions of the MIDI drivers (within cubase device setup) for my keyboard)
I have asked over at the steinberg forum but that has not yet yielded anything helpful to me

Besides, this problem has persisted for me since SX3, Cubase 5 and now Cubase 6, and has been the same on all MIDI keyboards I've attached over those years, which have been USB and MIDI-interface attached (thru the 1010LT). Even the MIDI out on my friends' Jupiter and Nord stage do the same. It was also like this on my old MAudio Audiophile 2496 card.

So my question is: Has anyone managed to get this working correctly ( in cubase specifically) even when changing overall projects latency (by adding plugins / changing buffer size)??
ie. so that the notes appear in the sequencer as they sound, not when they were struck?

There is always a chance that for all these years I've been setting it up incorrectly - but I know my way round these things and it has baffled myself plus some very clued-up music tech demon friends for a very long time. Please put us out of our misery!

Thanks in advance!

Re: Recording MIDI in Cubase with UAD plugins

man i had this problem...for years...and years...and it drove me completely nuts...and i tried every setting and every driver and every combination etc etc and it just stifled my creativity and pissed me off completely. i became so frustrated with the constant shifting of the midi parts i had played in and their constant lack of consistent "lag" i ended up not using my midi keyboard as much, and my music suffered. i did soundcard updates, cubase updates, firmware updates, OS updates, supposed registry fixes...the list goes on. There are pages and pages of suggestions for fixes for this problem with Cubase on the net and none of them work. or if they work, its until you reboot or it goes away for a week and then its back

i fixed the problem.

how?

1:/ updated my old duo core to the latest and fastest 6 core motherF%er CPU and kickass Intel motherboard i could find.
2:/ updated to Windows 7 x64 bit with 12 gig of the best quality memory i could buy.

problem solved. never had this problem again.

stop punishing yourself and update your pc and get off XP. you wont find a solution bashing your head against the wall as much as its meant to work with what you have...upgrade, get the problem outa your hair and move on with creating great music just spend the money man you will not regret it.

Cheers!
Blackout

Re: Recording MIDI in Cubase with UAD plugins

Thanks for the reply!

I set up Windows 7 on a spare drive, installed C6 on it and tested the midi timing. The notes still appear early in the sequencer!
I have re-tried all the "system timestamp" and "emulated port" options, but these made no difference.

Any more ideas?

Re: Recording MIDI in Cubase with UAD plugins

yeah ...well its the combination of a faster cpu and win7 that makes the difference. not just software man. i also had the problem on an Asus mb ....and when i upgraded i changed to Intel MB for that exact reason, to change everything to new to kill the problem. and it worked.

so maybe time to try intel or gigabyte or some other MB

Re: Recording MIDI in Cubase with UAD plugins

Blackout said:

yeah ...well its the combination of a faster cpu and win7 that makes the difference. not just software man. i also had the problem on an Asus mb ....and when i upgraded i changed to Intel MB for that exact reason, to change everything to new to kill the problem. and it worked.

so maybe time to try intel or gigabyte or some other MB

Click to expand...

But I do not have these MIDI problems when recording into Reason on the same system. So how can it be the hardware?

Re: Recording MIDI in Cubase with UAD plugins

because Cubase puts different stresses on the CPU and MB than other programs like Reason. And clips things like drivers earlier. it connects to hardware at a very low level, in a different way to other programs.

Cubase is a dog man. i mean...its a real pain and just runs with permanent bugs in it deeply rooted that they just dont wanna fix. But 5.5.3 is by far the most stable vers i have run. You just need to give it the right hardware. i would start with a different brand MB (ie new) and the fastest memory you can get running stable on it.

stop messing with software tweaks. focus on the hardware. yes i know it seems like the software is the culprit as Reason works ok but if you really wanna run Cubase then you need to choose hardware that will work with it and just bow to the Almighty Steinberg way of working.

if you dont wanna listen to me and think that the secret is somewhere in the settings (like i did for 2 years) then you will just go around and around and get no-where. Pull out your wallet and do some real fixes to the problem man

Re: Recording MIDI in Cubase with UAD plugins

what a load of nonsense in this thread! ..

the REAL (cubase) problem is, that when you are playing a vst instrument LIVE, your performance will be quantized to the BUFFER SIZE, yet recorded (and hereafter reproduced) "as played", since cubase only sends live-midi to VSTs during buffer "updates" .. other software might do it differently, but this is the root of the problem, and the only solution is to lower your buffersize .. (edit: this is only an issue with VSTi - not with midi-out to a hardware synth, ect).

.. but another thing to look out for is Windows MIDI vs. DirectMusic MIDI drivers. unfortunately there is no rule of thumb; this is down to your particular setup. a little troubleshooting will reveal a lot of things .. have a look at ftp://ftp.steinberg.net/Download/Ge...Port_Filter/midiportfilter_en_old_version.pdf .. MIDISPORT 8x8 is listed as a problematic interface, and since both Oxygen and MIDISPORT are m-audio products you might be a victim of said issue.

for what it's worth esi's m8uxl - using the ploytec drivers - is the tightest midi-interface around these days (i've tried midex 3, midisport 4x4, rme raydat's internal midi). i run a very elaborate hardware synth setup on this, and i have nothing but praise for it ..

Re: Recording MIDI in Cubase with UAD plugins

Cubase uses a "buffersize latency compensation" technology when recording, wich puts the recorded sound/midi into the projects more or less exactly how it was played. So if you use a buffersize of 1024 and play ahead of time, cubase will notice and sync the recorded midi ahead of the song just as you played it.. As far as I know there is no way to deactivate this "buffer compensation", but if anyone knows a trick or two i would be glad to hear it

Re: Recording MIDI in Cubase with UAD plugins

Well it looks like a lot of users are affected by this.

Judging from the info at the SteinBorg forum plus extensive personal testing, I now know the following:

- the problem exists with and without UAD plugins
- the problem exists when using NO MIDI INTERFACE to input the notes - ie. if you input the notes using the cubase on-screen QWERTY keyboard, they are still early. This eliminates all klnown MIDI interfaces.
- it happens even when there is NO VIRTUAL INSTRUMENT set up for the MIDI track. (can be tested using a microphone to pick up the sound of the mechanical keystroke) This rules out all virtual instruments.
- it happens on both INTEL and non-INTEL boards. This rules out the theory that "it must be genuine intel or it won't work properly"
- it happens even when using the Steinbeg-recommended MIDI settings in cubase (system time stamp, emulated port filters)
- it happens even with NO SOUND CARD in the machine (and no sound driver is loaded). This rules out all audio interfaces.

So all these things can be eliminated from the enquiry. What is there left? Cubase or Windows

My money's on Cubase. It used to work fine before they introduced the Plugin Delay Compensation stuff around version SX3, then when it went haywire Steiny told us we all suddenly had the wrong motherboard chipset / OS / MIDI / VSTi / USB interface / USB cable(!) / Keyboard / Audio card / haircut / shoes /drumsticks

I've sent several emails to Cubase support but they tell me they "can't reproduce the problem - and we have a lot of computers at Steinberg". I'm happy for them.
What useless pile of shite Steinberg support is! Basically the name of the game is to blame every one and every thing under the sun except Cubase, thus wasting a lot of people's time and energy in the the process.

Maybe somene should tell them that the software is supposed to be designed around the hardware, not the other way around.
They should fess up and fix it, instead of being obsessed on how to make the next cubase release look more like Dance e-Jay.
Not going to happen any time soon though, and I'm all done trying to figure it out.

(If I could find the "banging head against wall" smileys , this is where I'd put a really long row of them ---> )

The latency compensation is actually very useful when you do recording without using virtual instuments and fx inside cubase. example: recording a real guitar with a real amp with hardware fx while listening to a recording through cubase. But for midi recording this isnt helpful at all. Having said that, I dont see it as a very big problem since you can lower the latency by setting the sample buffers below 128. Most of us cannot notice much difference when latency is less than 10ms.

Aww, this problem has been around for ages. And it does drive me nuts. Yeah yea yeah yeah you can lower the effect by working small (64) buffers etc. but it's still not good enough. When you 'know' you had a great groove and timing playing down a part and few moments later playing it back only to notice it's 'almost' there but need to tweak the timings.

Try to explain to a great musician why my computer/daw is fugging up his/her performance. Last session I recorded keyboards both audio and MIDI. I've done this with VI's too with Cubendo's internal routing. I'm running very well performing 3.68ghz i7 setup, RME Digiface and Midex8 MIDI interface. Doesn't matter what MIDI input (RME/Midex/Novation) and emulated or not and with or w/o timestamps etc... It just doesn't work.

On the face of it, "always use the lowest possible latency" sounds like good advice, but in fact it's missing the whole point!

It's a bit like this :

Patient: Help me Doctor, it hurts when i lift my arm up.
Doctor: Don't lift your arm up then. What are you, some kind of idiot?

Turning all the 60+ UAD plugs off in my mix does indeed cause the latency to go down. However, my mix now sounds like total shit, which makes it harder/impossible to get into the vibe of the track when jamming-in some new MIDI parts (eg adding some percussion to a well-developed mix). I can play with all that latency no problem (lots of experience playing in large venues with dodgy monitoring) It sounds great live - but when I play back what I recorded, it sounds nothing like it did before.

Also what the "low latency" advocates don't realise is this: Even at the lowest latency settings, the MIDI is STILL EARLY. It's just a bit harder to detect (but it's there and still needs to be corrected to keep the groove locked in nice n tight).

I don't think they'll ever fix it at SteinBorg until they acknowledge that the problem actually exists - and we're still a long way from that happening. Even if you can get past the rather ridiculous forum trolls there, the support team don't want to know. How totally infuriating!

(In fact I have my own theory that the trolls over there are actually SB employees, actively trying to put people off from seeking support. They troll and troll and fire abuse at you until eventually some poor, frustrated assistance-seeking user snaps at them and criticises their ridiculous "answers", at which point the forum mods appear and come down like a ton of bricks on the user for "trolling", while at the same time completely ignoring the real trolls and their pages and pages of deliberately inflaming comments and mis-information. I've witnessed this many times and been on the receiving end of it myself several times. "Keep everyone confused", seems a good strategy for SB. At least then you can't blame them directly)

Oops sorry for the rant. 3... 2... 1... you're back in the room.

theres a big posting on this problem at the moment on the Cubase Forum...its 8 pages long now...lots of ppl complaining, no fix yet in sight

https://www.steinberg.net/forum/viewtop ... 19&t=17949

I have been following that thread. As usual, no actual input from Steinberg.
Everyone there is completely ignoring the fact that the problem persists even when no midi or audio interface is used, and that other DAWS (ableton, reason) do not have this problem.
Therefore, there is very high probability that cubase is broken, and needs fixing.
My own personal opinion is that SB know that it is broken, but don't know how to fix it.

I think the users who say they have no problem, actually DO have the problem, they just cannot detect it for various reasons (maybe they don't use high latency plugins like UAD)
I have told this to SB support, they tell me it is working fine on their computer.
So that's OK then!

Re: Recording MIDI in Cubase with UAD plugins

Hey blackout

Awesom-o said:

- it happens even with NO SOUND CARD in the machine (and no sound driver is loaded). This rules out all audio interfaces.

Click to expand...

There was a typo in the statement (my fault) at the time it got copied/pasted to the Steinberg forum. The erronoeus version mentioned recording with a microphone - which is obviously not possible with no audio interface installed!

I have now corrected the original post and would like to clarify it's very important meaning:

it is indeed possible to record MIDI with no sound card in the machine - just record the MIDI and VSTi output simultaneously (via a group into an audio track) while hitting a few random MIDI notes in realtime. Then zoom in and see the difference!

This proves it is not the sound card or driver to blame.

I'm sure that grey kitten will disagree though (you know he works for SB, right? ).

Now Steinberg say they will address this in the "next major release"!
Ahh, feel the vindication.

(And still the SB stealth-mod continues to troll and make pointless statements, causing as much smoke screen as possible, in order to hide real issues from public view )

Awesom-o said:

Now Steinberg say they will address this in the "next major release"!
Ahh, feel the vindication.

(And still the SB stealth-mod continues to troll and make pointless statements, causing as much smoke screen as possible, in order to hide real issues from public view )

Click to expand...

lol, are you referring to conman, whose only input to the forum is that if it doesn't work for you, it is you and couldn't possibly be the software?

The problem of early midi is (allegedly) fixed in Cubase 7, which came out yesterday.
Anyone tried it?

jmcecil said:

Awesom-o said:

Now Steinberg say they will address this in the "next major release"!
Ahh, feel the vindication.

(And still the SB stealth-mod continues to troll and make pointless statements, causing as much smoke screen as possible, in order to hide real issues from public view )

Click to expand...

lol, are you referring to conman, whose only input to the forum is that if it doesn't work for you, it is you and couldn't possibly be the software?

Click to expand...

Yes. He's one reason I deleted my account there. Little gobshite.

Awesom-o said:

jmcecil said:

[quote="Awesom-o":188qbu6m]Now Steinberg say they will address this in the "next major release"!
Ahh, feel the vindication.

(And still the SB stealth-mod continues to troll and make pointless statements, causing as much smoke screen as possible, in order to hide real issues from public view )

Click to expand...

lol, are you referring to conman, whose only input to the forum is that if it doesn't work for you, it is you and couldn't possibly be the software?

Click to expand...

Yes. He's one reason I deleted my account there. Little gobshite.[/quote:188qbu6m]
It would be more efficient just to put him on ignore :lol: Heck I can be annoying on that board (probably every board) too. Just point out your problem and click ignore on the noise posters. To be honest if someone posts an ORIGINAL problem, or question they usually get a reasonable response eventually.

Anyhow, it does look like it works so far.