Hi all,
I am having a problem with midi and I wondered if you could help. My midi seems to record early and am really struggling to resolve it. I know that this topic has been discussed at length in other forums but am having great difficulty in making sense of all the different and potential causes. I have had this same problem with Cubase 6 whilst operating on two different computers, on two different operating systems, on two different sound cards, and on two different midi keyboards. And still the problem persists. I failed to resolve it on the old computer but am keen to solve it on the new one. When recording, the midi initially came in doubled up in very quick succession. This was resolved by disabling one of the midi inputs. This solved the doubling problem. However the notes still come in early. I have tried to troubleshoot by doing the following tasks:
Sent my computer into Inta Audio for hardware testing
Updated drivers on windows, Cubase (to 6.5.3), midi device, and soundcard,
Tested different latency settings (am currently settled on 128)
Tested two different keyboards via midi and usb
Testing different midi settings in Cubase - ‘System Timestamp’ on/off
alternating between different midi inputs
moving the enable/ignore filters in the Steinberg programs folder
Activating ‘Constrain Delay Compensation’
Returning to an earlier snapshot of my operating system and testing Cubase midi. This allowed me to test Cubase whilst it was operating less VSTs. I heard these could be the cause.
After all this I still seem to get the midi coming in early. Although it seems to get slightly worse on the later version of my system (this contains more Vsts). Also can VSTs cause this problem even if you don’t have them running in a project but simply installed? The whole thing is driving me insane. I am planning to test Ableton and if this problem does not exist there, then I will move unless I can resolve Cubase. I have heard Ableton does not have the same problem. Please, please can someone help!
Thanks for the response. Below is a response to those details you picked out:
Hardware test: not entirely sure although was told that it was tested for potential midi problems. Nothing found.
Soundcards: old - Edirol DA2496
new - Roland Quad Capture
Computers: Both Desktop
OS: old - XP
new - Windows 7
Midi keyboards: M-Audio Radium 49
old M-Audio keystation
Yamaha SY85
Midi device - I have tested all keyboards via both midi cable into the Roland Quad Capture and via usb using the relevant keyboard drivers.
Midi settings Cubase: I tested different driver inputs that Cubase uses. For instance, when the Radium was connected via USB I tried activating these settings independently: M-Audio Radium ‘directmusic’. M-Audio Radium ‘Windowsmidi’. I tried both these with ‘Timestamp’ on/off. I then connected the keyboard via midi and tried activating only the ‘Quadcapture’ driver.
I notice that you highlight the discrepancy over whether the midi does actually come in early or whether it is my playing - is the Plonk test the best one for resolving this problem? Do you know of a good test for troubleshooting midi sync issues? Inta Audio mentioned that VSTs can cause the problem as their dlls are always loaded in the background of the program even if not actually loaded in a project. Can you shed any light on this? Your help is much appreciated.
Unless you know the exact behaviour of your MIDI device (MIDITest) and the exact latency figures of your soundcard’s safety buffers, buffers, FW/USB clocks, ADC and mic distance (in other words: half the roundtrip latency), the plonk test is useless.
Because when one wants to compare MIDI with audio you have to do some math to compare them and need to know the numbers.
I would suggest to first run the MIDI test to see if your MIDI devices behave within the MIDI specifications.
Or start with a MIDI i/o loopback in Cubase and see how things line up.
What quantise setting and what sample rate?
Again.
And what are your VST settings in Devices Setup?
Quantise settings were strangely danced about in the other threads too. Leading me to suspect either bogus threads or plain old incompetence. I stopped wasting time on those as they couldn’t answer simple questions.
Hey Dom, thanks for the details. I was just afraid you wouldn’t be taken seriously without a better description of your setup.
When you say early MIDI, what are you comparing it to? A pre-recorded audio track? The Cubase metronome? A MIDI drum track? If you are playing along with the metronome is it an audio or MIDI click?
And how much early are you talking about? 1ms? 10ms? 50ms? Please don’t say a 1/16 note and not give a BPM.
Hopefully this topic won’t go in a million different directions with no conclusion like the other similar ones have gone lately. I’d really like to see somebody solve one of these “early MIDI” problems.
As far as the internal MIDI workings within a given system go I wish I could offer you some more help. niles (and some others around here) are much more educated in that area and should be able to explain things better than me.
All I know is I’ve never noticed anything funky in this area (like the majority of other users) so it’s not a general Cubase problem or this forum would be blowing up. More likely a hardware incompatibility issue. What motherboard are you using?
We know there’s an ongoing problem but until the “victims” start giving meaningful info it can’t be fixed via a forum. It took weeks of mental blackmail to get any of them to give any basic system specs.
Any pertinent enquiries seemed to be met with abuse along with suggestions that anyone enquiring in a sensible direction (other than joining in the “Why don’t they fix it!?” chorus) didn’t know what they were talking about. Very curious.
I do hope this isn’t just a disguised continuation of the same confused guff.
I am having the same problem too. i would say that I am getting notes appearing about 34ms early on the cubase time line. I hope people can help Dominixique identify the source of the problem - in a nice non-adversial way. I’ll see how this thread goes before distracting it with my set up.
Give setup details or it doesn’t exist.
Please distract us from the mysteries of an anonymous system with a vague problem no-one else on the net has except in this forum.
Thanks all for the responses. I have mentioned most my specs in the second post I wrote in this thread. In addition I am running Cubase on an Intel Core i7 2600K CPU 3.40 GHZ. After going back to an earlier image of my OS, with no vsts installed, I found that the problem is better. I have reinstalled some of my plugins and the midi in Cubase remains solid. I am gradually reinstalling them all step by step, to see which, if any are causing the problem. I suspect it might be the Waves plugins but I will see. Can VSTs cause the problem, even if not used in a project? Remember some plugins such as waves install the Windows C++ 2007/2010 drivers as well. I will inform this forum if/when I find the problematic VST.
In terms of my current recording results. When I record midi and audio in from an external keyboard they have a time difference of about 1ms or less so pretty good. When i record midi in on a VST instrument I play a lot of notes in early but that must be me pre-empting things unless the test I just mentioned is not full proof. Before I re-installed windows and cubase my midi was terrible. Notes came in consistently early and note lengths were altered randomly. Thoughts? Can VSTs cause this problem, either when used in a project or not? Is the test I mentioned at the beginning of this paragraph full proof?
That’s very interesting information for people suffering MIDI placed ahead in the project.
It makes one suspect an additional service, drivers, application or even setting is interfering with the system’s event timers.
Did the rollback used the exact same device drivers (chipset, HPET, ASIO, MIDI) as the failing instance of the OS?
I doubt an unloaded VST will have any effect on MIDI placement. In fact I even doubt a loaded VST will have any effect on the actual MIDI data stream, unless it stresses your system of course.
Keep us posted, you’re the first who manages to turn a system suffering from the infamous early placed MIDI data, to a properly behaving one.
Must be said, you’re also the first one that totally rolls back and troubleshoot the way it should. Step-by-step.
Thanks for more detail. Still no sample rate and quantise I see (4th time asking?)
Still curious how basic queries are danced around. Sample rate is set in the Soundcard and quantise is set in the Cubase Toolbar.
If you’ve improved it by a goback then it will be something interfering with the timing of the soundcard driver in the Windows update.
Your midi is early in relation to what? The audio or the Cubase timeline or both? As the audio isn’t quantised how can you be sure that either your playing or the way audio and midi is sent by the system sends it to the point indicated on the timeline. ie: the audio is slightly late and the midi quantised to the nearest quantise point on the Cubase timeline and so appears “early”?
If recording from point zero (00-00-00-00 seconds) in Cubase then does your first midi note not get recorded?
Or do you work with an offset?
Please do the plonk test and the “record click” test on your (faulty) system, so the problem can be made more transparent.
The plonk test will show you the amount of time your MIDI is early (by comparing the placement of the MIDI to the placement of the audio “plonk”).
The “record click” test will show you if eventually your audio is late (as opposed to or in addition to your MIDI being placed early). Just close mike your speaker and record your Cubase click (or a MIDI generated click), then compare the audio placement to the grid to determine if it is significantly behind the grid (which it probably won’t be).
The plonk difference would then be your negative MIDI offset. If it’s significant, then yes, you’re experience the so-called “problem Nr.2”, which seems to be a true (possibly hardware-related) Cubase issue.
No-one doing the “plonk test” (white wine could be better for this) ever reports their quantise settings or their sample rate so if you could do this it would be a first.
And if one does the “plonk test” using a count in and no bar off set the first midi not disappears? (I do look forward to the great silence on that question)
Sceptical? Me? You betcha.
Have not had early midi but using a Roland OctaCapture usb device I find you do have to be careful with the driver settings and check them fairly regularly to see if they work as they should. Sleep mode seems to disagree with them for one so I’d check the Roland site for the latest drivers and firmware for that QuadCapture.
Ditto for any other usb devices and check audio forums for mention of any problems similar to your own.
Hi all. After installing the majority of the plugins initially installed on my previous OS, I still have very accurate midi timings. When I perform the plonk test I get a difference of less than 1ms between midi and audio. In terms of my settings that Conman has been mentioning, I am running at 24 bit and a sample rate of 44100. I am not sure about quantise settings. I am not using quantised snap to grid when I record but I imagine it is something other than this.
The key to resolving the midi early problem was to re-install my OS, and having a Acronis image backup of my original C drive, to return to if needed (my Cubase projects and samples are on a second and third hard drive). Re-installing windows and updating the drivers for windows, soundcard, midi keyboard, cubase, and vsts, has worked in overcoming the midi issue, for now anyway. When I initially started testing the midi on the re-installed OS I panicked as I still thought the midi was coming in early when testing against the Cubase metronome. At times I even thought the midi was coming in irregularly because the odd note was on the desired grid line or a tiny bit late. The plonk test enabled me to eliminate the view that the midi was recording erratically as the midi notes were following the audio exactly. Subsequently, when I record midi along to the drums of a track, the notes record very accurately. I therefore, must have been playing notes to rapidly (problem category 1) and panicked as a result. The plonk test was vital in concluding this.
On my previous OS the midi timings were mostly early but also erratic and the note lengths were often altered. Sometimes there would be a pile up of notes with no quantise activated. When I returned the computer to manufacturer, the guy testing mentioned how horrible the midi timings were, so I was not hallucinating. He suggested a fresh install after testing the hardware.
Anywayz I would advise anyone who has midi sync problems to try a clean install of windows and update everything else as when I tried to resolve this problem on the previous os it never seemed to work (I tried all the common troubleshoots). The problem in fixing from within your present setup, is the large array of potential offenders and locating which one is causing the issue. The VSTs were not the cuprit as I tested the midi after installing most of them and the timings were accurate. I have also found that connecting the keyboard via midi through the sound card, rather than via USB (that uses the midi drivers), works best. The original problem was very likely caused by a driver clash between the sound card and windows. This may seem a cop-out to those looking to resolve the problem from within their current operating system but it worked best for me. This may all change tomorrow and I will be begging for advice again. Anyway, thanks for the help and I hope this can be of some use to others experiencing the same difficulties. Much appreciated.
Great to hear you’re not experiencing the problem any longer.
As to Conman’s brilliant remark about quantise settings: noone else (besides himself) understands that point either, since anyone with half a brain would turn auto quantize off when recording, as you did. Yet, he keeps asking for some “quantise setting” over and over again for reasons best known to him.
I have been struggling for weeks with mine since this thread stopped. Notes continued to appear about 36 milisconds early on the time line making my playing sound sloppy. However this evening I seem to have fixed my timing issue now so thought I should share in case anyone experiences the same cause as me. I have a Yamaha S90 and found that Yamaha have recenty introduced updated S90/Motif USB/midi drivers for windows 7/32 bit and various other mac/pc setups. The update is dated August but I presume released more recently as I have searched before.
I had to turn off (untick) ‘use system time stamp for windows midi imports’ in the devices/midi port setup for the problem to go away. Now i know the problem is fixed because some notes are slightlty early, some spot on time and some slightly late - rather than all very early!
