MIDI recording late

[bayswater]

I tried this out with a simple project with 2 MIDI tracks, one with output to 4Front EPiano, and the other to nothing. Turn on Multi-Record. Set to record and play a scale. Go to MIDI editor and zoom in to the sample level. Both sets of MIDI events are at the same time -- exactly. Repeat with section MIDI track set to a BCF2000. No change. Set it to a hardware drum machine. No change.

So this has to be specific to DP 11, or some version of CoreMIDI, macOS, etc.

[smashprod]

That wouldn’t make the bug happen. Record a MIDI track to click (quarter notes) with no instrument tracks and no VIs anywhere in the sequence. Be as precise as you can. Then add a new instrument track with a VI, and a second MIDI track. Assign that track to the new VI. Record quarter notes to click on that second track. It will be later than the first one, although you cannot measure precisely because you’re recording live.

Reproducible on two different Intel machines with different versions of Mac OS and no common KEXTs or background processes. Only common elements are DP 11 and Intel.

[bayswater]

So the critical difference is whether a VI is present in the project, not whether a MIDI track is using it?

[smashprod]

Yes, whether it is present.

[Kurt Cowling]

So, I just tried this on my Intel iMac. I found the MIDI timing was good until I actually assigned the MIDI track to the VI. Having the VI present didn't affect the timing of a track that was assigned to my S90-ES. I wonder if the previous result (indicating it just needed to be present) was from having both MIDI tracks in record at the same time. I left the MIDI track assigned to the VI out of record.

BTW, I was using "built-in" for audio hardware and checked the timing by recording the noise of me hitting the keys as reference.

--Kurt

[bayswater]

Got it ! That's really odd. Doing that way, I get an average offset of 46 msec. With the MIDI Patch Thru sync off, the delay increases to about 70 msec.

Good catch.

My templates always have a VI, so I'd never have noticed.

[smashprod]

A little off-topic: I am readying myself for the eventuality that there is no fix for this. Is anyone using DP11 on an M1 Mac Mini for professional scoring (large templates)? I would run all my VIs on a slave, it my feeling is that 16GB is still not enough to run a massive DP template.

[HCMarkus]

A little off-topic: I am readying myself for the eventuality that there is no fix for this. Is anyone using DP11 on an M1 Mac Mini for professional scoring (large templates)? I would run all my VIs on a slave, it my feeling is that 16GB is still not enough to run a massive DP template.

That would be future me (M1 Pro or Max when MiniPro is released). Left my crystal ball at the beach, so I can't provide further details.

[bayswater]

Why not run the massive template on your existing setup and look at how much RAM it uses?

[smashprod]

Good idea!

[HCMarkus]

One thing to keep in mind when using sample-based VIs, users can adjust the pre-load size (or whatever it is called - where the VI places the first bit of every note into RAM to allow immediate playback before streaming from disk begins).

With the super-fast M1 SSD, we should be able to minimize pre-load and use the smallest amount of RAM possible.

[smashprod]

More info on this bug; Matt at MoTU was able to reproduce the exact same problem (MIDI recording late with high buffer settings and VIs) in Ableton using an Intel Mac. So it's not only a problem in DP. OTOH, Pro Tools doesn't have this problem.

I'm still trying to find out whether the higher-ups at MoTU are aware of this and if there's a workaround or a fix. If not, I have to switch to an M1 Mac, or find a competing DAW that doesn't have the problem. Don't known what that would be, DP is so geared for scoring, but I guess a lot of composers use Logic. Dread the learning curve though; jumping ship on DP is my last choice...

[James Steele]

This is interesting. I finally got around to testing this and had similar delay on the VI track. I turned off "Sync Recorded MIDI to Patch Thru" and on my system it seemed to be much better. I was only recording quarter notes to the VI track at 120bpm.

This may be totally off base and irrelevant, but I recently connected up my old Oberheim Xk master controller keyboard. This is so old it does NOT have a USB connection so I have to connect it via a MIDI cable to an input on my MTP/AV. This makes me wonder if there is any difference between being connected this way and USB? I'll have to hook up my Axiom 49 via USB and compare results.

[smashprod]

I've done both MIDI cables and USB-- no difference.

[James Steele]

I've done both MIDI cables and USB-- no difference.

Well there goes that theory.