MOTUnation.com

Hi,

I've noticed some poor performance in DPs MIDI timing. I know MIDI is not a great standard, being serial etc but I'm hoping to tighten things up in terms of timing. I'm using both the MIDI in/out on my Focusrite Clarett 8pre and a MOTU Fastlane.

Thanks in advance.

Darren

PS Yes I can play to a click!

Make sure you don’t have any latency inducing plugins on the Master Fader in VI tracks while recording MIDI.

dazzjazz wrote:Hi,

I've noticed some poor performance in DPs MIDI timing. I know MIDI is not a great standard, being serial etc but I'm hoping to tighten things up in terms of timing. I'm using both the MIDI in/out on my Focusrite Clarett 8pre and a MOTU Fastlane.
!

Can you be more specific about the problem you're having? Are you saying that when you record your MIDI takes, the notes are not landing on the grid where you think you played them?

Or are you saying that even if something is quantized strictly, that during playback you can hear MIDI slop happening?

Are you talking about external instruments or software instruments?

MIDI over USB 1.1 is sample accurate. Something’s wrong but, without knowing more about your setup, I haven’t a clue.

to be fair about the accuracy, MIDI over USB will only be sample accurate if the MIDI device supports MIDI timestamping from CoreMidi. Not all do and its hard to find out for sure in most cases.

CoreMidi itself includes MIDI timestamping, but not even all apps are fully aware of it. Its a hit or miss thing.

I think if you use MOTU MIDI interfaces, you're more assured of MIDI timestamping doing what its supposed to to do, with DP being the sequencer involved.

The OP can try to ask the manufactures they are using to see whether the MIDI implementation is compliant completely with MIDI timestamping all the way to the device.

Actually, even MIDI timestamping is not "sample accurate". What MOTU has said in the past about MIDI timestamping accuracy is that its sub-millisecond accurate.

Your software instruments will render MIDI sample accurately, but MIDI sent to external gear is not sample accurate at all...but can be accurate enough.

There is this one sample accurate MIDI interface, kind of clever and interesting....

https://www.expert-sleepers.co.uk/usamo.html

The biggest problem with most MIDI interfaces is jitter. The discontinued Roland's have rated the best for that, and MOTU a close second place. But ultimately I think using a MOTU interface with DP is the way to go, sub millisecond accuracy through MIDI timestamping. USB on its own does not provide that. The device itself needs to have a clock that can timestamp each MIDI event before making it available for the driver to receive via CoreMidi and hand over to the ap with the timestamp. A lot of devices do not have that.

This old thread from years ago may help with insights about MIDI jitter problems.

http://forum.cakewalk.com/MIDI-quotJitt ... 78006.aspx

dewdman42 wrote:
Can you be more specific about the problem you're having? Are you saying that when you record your MIDI takes, the notes are not landing on the grid where you think you played them?
Are you talking about external instruments or software instruments?

That’s right, MIDI notes are not landing on the grid where I think I’ve played them. I have to play ahead of the beat in a really unnatural way. This happens with software instruments and external instruments.

Besides potentially being an issue with the unintuitive "Sync Recorded MIDI to Patch Thru" setting in MIDI patch thru...

There's been a MIDI recording accuracy bug in DP for 3years since post v9.1. Likely a bug from the pregen enhancement. This has been confirmed by motu. Not sure it was fixed in v10.x. Was not in v9.x. Simply put, the placement of a note-on event would shift relative to your audio interface buffer setting. The lower the more accurate, the higher, the less. This should not happen, it should be compensated for and independent of audio buffers. This was with a 1248. Again, confirmed by motu tech & dev. To me this is a P0 bug in severity. Others have posted about this issue in confusion. There may be a lot DP users scratching their heads thinking the timing is them...it's not.

as a temporary work around can the recording offset be used to make a manual adjustment?

mwilloam wrote:There's been a MIDI recording accuracy bug in DP for 3years since post v9.1. Likely a bug from the pregen enhancement. This has been confirmed by motu. Not sure it was fixed in v10.x. Was not in v9.x. Simply put, the placement of a note-on event would shift relative to your audio interface buffer setting. The lower the more accurate, the higher, the less. This should not happen, it should be compensated for and independent of audio buffers. This was with a 1248. Again, confirmed by motu tech & dev. To me this is a P0 bug in severity. Others have posted about this issue in confusion. There may be a lot DP users scratching their heads thinking the timing is them...it's not.

I confirmed buffer changes MIDI note placement in v9x. I, too, wonder if this is fixed in v10?

Wow, interesting info regarding the bug and its confirmation my MOTU.
Really needs to be fixed.

dazzjazz wrote:Wow, interesting info regarding the bug and its confirmation my MOTU.
Really needs to be fixed.

That said, perhaps you might try changing the buffer size you use when recording... you may find the sweet spot where things line up really well for you.

And, btw, I am not certain this irregularity still exists in 10. I haven't tested.

Interestingly, I am experiencing no MIDI lag
with hardware in DP 10.1. In 9.52, any MIDI
I recorded on external synths would play
back anywhere from 15 to 30 ticks late, so
I would have to shift the MIDI earlier, or shift
the recorded audio earlier by that amount
to get it to line up with my sequence properly.

The issue seems to be gone in 10.1. No MIDI playback
lag with hardware.

David Polich wrote:Interestingly, I am experiencing no MIDI lag
with hardware in DP 10.1. In 9.52, any MIDI
I recorded on external synths would play
back anywhere from 15 to 30 ticks late, so
I would have to shift the MIDI earlier, or shift
the recorded audio earlier by that amount
to get it to line up with my sequence properly.

The issue seems to be gone in 10.1. No MIDI playback
lag with hardware.

Just to be clear, the issue I spoke of was never playback lag, but rather where the MIDI note event was recorded relative to when it was actually played in time. MIDI penciled in on the grid played perfectly. MIDI recorded, technically played back perfectly...problem was it was playing back inaccurately recorded MIDI.

Easiest way to test is setup a mic, close as possible to a MIDI keyboard key(or splice a MIDI->ts cable if you know how). Create an audio and MIDI track. Start with audio buffer of 64. Hit the key hard and quick for a solid transient. Record the transient from the mic and the MIDI note on. Zoom the tracks and observe any transient/note offset. Retest for audio buffers of 64->2048. No matter the offset it should consistent, independent of the audio buffer. If it changes based on the buffer, that's the bug.

mwilloam wrote:Just to be clear, the issue I spoke of was never playback lag, but rather where the MIDI note event was recorded relative to when it was actually played in time. MIDI penciled in on the grid played perfectly. MIDI recorded, technically played back perfectly...problem was it was playing back inaccurately recorded MIDI.

Easiest way to test is setup a mic, close as possible to a MIDI keyboard key(or splice a MIDI->ts cable if you know how). Create an audio and MIDI track. Start with audio buffer of 64. Hit the key hard and quick for a solid transient. Record the transient from the mic and the MIDI note on. Zoom the tracks and observe any transient/note offset. Retest for audio buffers of 64->2048. No matter the offset it should consistent, independent of the audio buffer. If it changes based on the buffer, that's the bug.

+1