MIDI loop picking up note from previous part

Issues with MIDI seem to be coming thick and fast for me at the moment…

I’m looping a MIDI part that starts on beat 1 with a longish note but when playback loops back round it plays back very short.

The problem is the previous part: its last note (the same as the first of the loop) ends at the end of the bar but is sounding briefly at the restart of the loop and masking the note that should play. If I shorten this last note even by 1 tick the loop plays correctly. If I then restore the note length the loop continues to play correctly. The same thing happens if I leave the note length running up to the bar end and instead shorten the part by 1 tick. Which I find a bit odd; I was expecting the note to play its full length and cause the loop to mis-play.

If I shift the pitch of the last note before the loop up by a semitone, instead of shortening it, the first note of the loop plays correctly but the short now discordant last note of the previous part is clearly audible.

In Preferences > MIDI, Extend Playback Range of Notes that Start before the Part is 0 (as advised by @Martin.Jirsak to a post with a similar-sounding problem I had in 10.5 and Length Adjustment makes no difference (I’ve had it down to -20).

I’ve tried the project in previous Cubases and the further back I go the weirder it gets: octave and volume shifts which must be due to project incompatibilities as I’m sure they never acted up quite like that at the time I was using them. I’ve failed to find the same problem in another project I’ve been working on in parallel but it doesn’t use much MIDI (or the Halion Symphony Orchestra for that matter, which itself is not entirely out of the frame). Anyway, my brain now hurts quite badly so I’m posting this and going to bed. I’m at a loss and it’s holding me up as I can’t currently trust what I’m hearing.

Night all,


Do you use any MIDI Insert on the track by any chance?

Are you on Mac or Windows?

Could you attach the screenshot of the MIDI Part to see the Start and the End, please? Is it reproducible?

Hi Martin, glad to hear from you. Hope the screenshot is okay. Just ask if you want to see anything else. As you can see, I’m on W10 and there’s nothing fancy on the track. As for reproduction, I’m working on that. It’s happening in other parts of the project, including other tracks using the same HSSE/HSO combination, so we can at least discount the events themselves. But it’s hard to reproduce reliably by, say, shortening a part to the end of its last note then butting the next part up against it.

There’s a bass track also using HSSE but I can’t get that to trigger the problem, in spite of being written in the same way, i.e. quantised notes. A memory is stirring that I have experienced this with a bass line in another project and that would have been similarly configured but I haven’t been able to find it.

Hope that all helps. Cheers for helping out.

I think I’ve found it. The last note before the loop is in fact slightly longer than the bar but only by 1 millisecond and so intrudes into the loop and over its first note. This is too small to be picked up by the Bars+Beats grid so the events looked lined up.

Quite how it came about I’m not sure but I did an ‘Insert Silence’ in the intro of the previous version of the project and the MIDI ended up substantially out - in other words, I messed it up. Off the top of the head, maybe because I’m using a Tempo Track so I’m inserting an amount of time which doesn’t accord with Bars+Beats at some point and things get out of line. Maybe I had the wrong Timebase setting. I remember having a good few goes at it, and being worried about milliseconds and bars/beats not quite matching.

Edit: in fixing the note in question, the end point should be .988 (from .989) but I can’t manually enter .988 - it rounds down to .987. This couldn’t all be the result of a rounding error, could it?

And sometimes the end point before the loop matches on milliseconds the start of the loop but still overlaps. Sensitive or what!

1 Like

And just to round the discussion off, you can also find left-over MIDI notes of length "0"ms but with a velocity so they sound. Presumably they have length of <1ms. Hard to spot…

What exactly is the accuracy of MIDI timing? Do you know, Martin?


I would expect it’s sample-accurate in Cubase. But I don’t know for sure, sorry.