MIDI recording late

[smashprod]

Here's an updated list of CPUs with the problem:

• 2012 12-core Mac Pro 5,1 ("cheesegrater")
• 2017 27” iMac 18,3 with Quad-Core i5
• 2017 Macbook with 2.9GHz Quad-Core i7
• 2018 Mac Mini with 32GHz Six-Core i7

The MoTU engineering team is being informed. Matt Batson says he can't give me a timetable but will keep me informed on what the engineers say. Keep testing and keep 'em coming!!

[dewdman42]

Thanks for your follow up and diligence on this. Hope MOTU will address it.

[dewdman42]

Ditto. On the 2018 Mini i7, this is easy to repeat in DP, but I'm unable after several tries to repeat it in Logic. Not that Logic timing is perfect -- after a lot of tests, it looks like Logic is consistently recording MIDI about 15 msec early regardless of the buffer size.

That could be because of your loopback IAC approach of testing....because the output of MIDI has different behaviors in LogicPro also. Are you using external MIDI tracks or the external instrument plugin in that test?

[bayswater]

Ditto. On the 2018 Mini i7, this is easy to repeat in DP, but I'm unable after several tries to repeat it in Logic. Not that Logic timing is perfect -- after a lot of tests, it looks like Logic is consistently recording MIDI about 15 msec early regardless of the buffer size.

That could be because of your loopback IAC approach of testing....because the output of MIDI has different behaviors in LogicPro also. Are you using external MIDI tracks or the external instrument plugin in that test?

That makes sense except I wasn't using IAC when testing this on Logic. I couldn't get Logic to respond to transport sync messages from DP, so I just did it all within Logic, using an external MIDI to a DM5, and Instrument tracks with the Logic drum VI.

[dewdman42]

you lost me on how you measured 15ms. Try the external instrument plugin by the way.

[bayswater]

you lost me on how you measured 15ms. Try the external instrument plugin by the way.

Record MIDI tracks to a metronome, one to an external hardware device and one to a VI about 30 bars of quarter notes. Dump the Event lists into a spreadsheet. Calculate that average difference in ticks using 480 ticks per quarter note at 120 BPM. With these parameters, a tick is close to a msec, but is easily converted.

It’s easier to do this if you can re-record an already produced MIDI track with quantized timing. When I did this to look at the delay in DP, the delay between no VI and a VI, the difference was within 1 msec for every recorded note. The difference varies a lot more than that if you play each of the tracks but over the course of 20 bars or so you get a consistent average difference from one test to another.

I don’t know what the point would be of using an external instrument plugin — what does that bring to the party? How is that different from sending MIDI directly to a hardware synth from DP or from Logic? AFAIK, this is just a means of recording a MIDI track without a VI in the chain. To that end I did not see any difference in timing depending on the destination of the MIDI track: an IAC, a device like a BCF2000, a hardware synth like the DM5, or anything other available MIDI destination. Only when the destination is a VI is there a delay in the recorded events.

[dewdman42]

I’m still not understanding what you did to determine 15ms early recording in logicpro. To me sounds like you have a lot of moving parts that makes it hard to determine anything definitive, if I could even understand what you are trying to explain. But anyway logicpro has a slider you can use to adjust the record offset in such a case.

If you aren’t using an iac loop now to do your test then never mind my comment about external inst plugin

[bayswater]

I’m still not understanding what you did to determine 15ms early recording in logicpro.

Start a session with one external MIDI channel. Play quarter notes to the metronome recording to this channel. BPM-120, buffer=1024.
Add an instrument channel, and do it again. Now you have two MIDI regions on two tracks. Export each.
Import these to DP, and print the Event List for each to PDF. Copy and paste the PDFs to a spreadsheet. Parse the data and calculate the number of ticks from the start of the sequence to each note for each track. Take the difference between the values for the two tracks and calculate the average. I get about 15 msec difference with the Intrument track being earlier.

[dewdman42]

for the external MIDI track use the external instrument plugin INSTEAD OF an actual external MIDI track. That is generally the preferred modern way to use external MIDI instruments in LogicPro. What I can tell you about using external MIDI tracks in LogicPro is that it was done a LONG time ago and the assumption at the time was that you'd be listening to your MIDI instrument through an external mixer and not with software monitoring...or something to that effect. The end result is that external MIDI tracks make different assumptions about exactly when to send the MIDI to the external MIDI instrument and also in a way that can affect the way you record the notes to the track (record offset).

The external instrument plugin is the modern and preferred to to deal with external MIDI instruments because it aligns external MIDI with the latency of not only your sound card, but also any plugin latency that may be in your project. I suspect that 15ms difference will go away.

But....

you haven't identified in this test with the instrument track test is early or late or not...your're just saying one way is 15ms off from the other way. So they are off from each other, but which, if any, is the correct one?

[bayswater]

for the external MIDI track use the external instrument plugin INSTEAD OF an actual external MIDI track. That is generally the preferred modern way to use external MIDI instruments in LogicPro. What I can tell you about using external MIDI tracks in LogicPro is that it was done a LONG time ago and the assumption at the time was that you'd be listening to your MIDI instrument through an external mixer and not with software monitoring...or something to that effect. The end result is that external MIDI tracks make different assumptions about exactly when to send the MIDI to the external MIDI instrument and also in a way that can affect the way you record the notes to the track (record offset).

The external instrument plugin is the modern and preferred to to deal with external MIDI instruments because it aligns external MIDI with the latency of not only your sound card, but also any plugin latency that may be in your project. I suspect that 15ms difference will go away.

But....

you haven't identified in this test with the instrument track test is early or late or not...your're just saying one way is 15ms off from the other way. So they are off from each other, but which, if any, is the correct one?

Thanks for the tip on the plugin. I’ll start using that where it might help. For the most part I use Logic only when someone sends me tracks in Logic or GB files, so I don’t know it that well, although I’ve had it longer than I’ve had DP.

On your question, the external MIDI track is closest to the beat when I look at the notes in the piano roll editor. The instrument track generally shows notes landing a bit early. So, the direction of the difference is opposite to what I see in DP, but it is much smaller.

[dewdman42]

use the plugin, so that the timing is consistent with your other tracks. Use the record offset slider (in audio preferences) to eliminate the 15ms.

[smashprod]

I've discovered that this timing bug is not consistent; in other words, the lateness varies. One average I was seeing about 40ms delay, but it's not an exact figure, and sometimes, for no reason, the lateness is much less.

There are way too many variables for me to test for. I am counting the days to Vienna releasing a Silicon native version of VEPro 7, then I can switch to an Apple Studio.

Meanwhile I am going to look at the latest update, DP 11.11. There is nothing in the release notes about them dealing with this, so I don't expect there is a fix.

[bayswater]

I did not get a consistent delay either when I was testing this. The delay, as far as I could measure it varies by +/- 40 msec from one MIDI event to the next. But I attributed this to the lack of any methods to get a precise measurement of the delay.

Have you learned anything about this in recent Version 11 updates? I’ve been deferring updating because there aren’t many new features of interest to me and so many odd bugs are being reported, but a timing fix could tip the scales.

[mikeharper]

Had the same problem with a very good RME card. I believe it was a driver/resource conflict between that card and my micro express. Went away when I got a MOTU 8A.

[dewdman42]

Yea I'm still getting same behavior as a few months ago, if the buffer is set to a large enough value, the recorded MIDI events are recorded late to the MIDI track by a large amount, presumably the same amount of time as the buffer. Kind of annoying, but at least if you use a small buffer to record it seems to be ok.