Notifications
Clear all
Topic starter
In a recent thread I sought advice and (with Bad Mister's extensive input) reached a point where my Motif XF8 was syncing well with my hardware audio recorder (Roland VS2000CD), with the audio recorder as MIDI clock master.
With my setup right now, if I record some MIDI drums on one track of the Motif sequencer in SONG mode, quantize it to 8th notes, then record the audio to my hardware audio recorder, I will find that the recorded drum hits end up just around 2.7 milliseconds after "the grid" on the audio recorder, consistent every take, and with very little timing jitter. I think this is absolutely excellent, considering all that has to happen to sync the machines, trigger the events and record the Motif's audio output.
However, one strange phenomenon remains: if I mute the sequenced drum track, and instead record the audio of the Motif's metronome, I find that the recorded metronome audio sits much later, in fact around 25 milliseconds after "the grid". Again, this is consistent every take and there is no significant timing jitter.
Since 25 milliseconds late is enough to be heard, I'd like to try and understand how this could be the case, and resolve it if possible. I'm writing to ask whether anyone might have an idea of what type of reason could be at the root of an issue like this.
Some setup info:
- The VS2000CD is MIDI clock master and also transmits SPP.
- Its MIDI output is connected directly to the MIDI input of the Motif XF.
- I use the Motif's SPDIF digital audio output as my input to the hardware recorder.
- I get the same result whether or not I use a pre-roll/lead-in period on the recorder.
- I get the same result whether the Motif XF is set to MIDI CLK sync mode or AUTO.
- I am not using any clock shift or other PlayFx on the Motif track.
- I have verified that the quantized drum notes sit in the Motif sequencer at beat 1:000, 1:240, 2:000, 2:240, and so on.
Additional tests done:
I tried substituting two other recording devices/MIDI clock masters in place of the Roland VS2000CD recorder. One was a well-known DAW that I will refrain from mentioning by name unless asked, and the other was my AKAI DPS16 hardware recorder. For the DAW, I used USB connection to the Motif to transmit the MIDI clock.
When using those alternative devices to send MIDI clock + SPP and record audio from the Motif, in neither case did I find the Motif's metronome audio being recorded late. In fact, the metronome audio recorded very slightly earlier than audio of the sequencer events, and in any case both were within 2-3 milliseconds of being "on the grid", ie. close enough for any reasonable purpose.
I recognize that this suggests there is no problem with the Motif XF and the problem likely lies with my other gear. I would still like to learn about possible causes, if you have any ideas.
Thanks for reading and for any ideas you may be able to offer.
Posted : 12/07/2020 1:28 pm
First thing I would look at is: what is different in how you are dealing with the drums and how you are dealing with the metronome. If the drum groove transfers are fine (no drift, tight timing), and the metronome is late by a significant amount (but significantly, again, no drift)... then how are you ensuring the start of the Metronome? Could be when you start the metronome experiment you’re starting it off the beat... it seems to hold the timing (no drift)
Try recording the Drums again, this time with the Motif XF metronome active, simultaneously. You realize the Drums, quantized, are using the same clock information as the metronome and they both can be active simultaneously.
In the Motif XF set the Click mode to “Always” or “Play” so that you can hear it while the drums play...
DAWs, can all handle clock sync. It’s a fundamental.
DAWs will time stamp the audio when it first arrives at the computer and position it properly according to the Input and Output delay. The computer ‘knows’ it cannot process it instantaneously, so it measures the delay, and good DAW software will precisely place the audio where it should belong.
‘Late‘ can be understood, but ‘early’ is an indication of a problem...An event should not occur before it’s supposed to exist! That’s overcompensation and is a problem. You can probably adjust that.
Posted : 12/07/2020 2:25 pm
Topic starter
Thank you Bad Mister.
I ran a test, as you suggested, with the sequenced drums and the Motif XF metronome both playing simultaneously, slaved to MIDI clock.
Setup
- I recorded to a stereo pair on my hardware audio recorder (tracks 7 and 8, linked as a stereo pair)
- I panned the drum voice fully left in the Motif voice mixer in order to have it recorded on only track 7 of my recorder.
- The metronome is centred, so its audio appears on both tracks 7&8. However, as it is the only sound appearing on track 8, this makes it easier to distinguish the timing of the waveform transients.
Under this test setup, I again found that the metronome audio appeared significantly later than the audio of the sequenced drums.
The following image shows track 7's waveform (a combination of both sequenced Motif drums + Motif metronome), with the red now line located at the start of bar 6. I've expanded the vertical scale somewhat to improve ease of viewing.
The following image shows track 8's waveform (Motif metronome only), at the same now time and zoom:
To reach the start of the Motif metronome transient, I need to move the red now line 21 ticks later (@108bpm). This is a little over 24 milliseconds. The SMPTE-style time display at the top shows the difference as equivalent to 74 subframes (from 16f66 to 17f40).
Additional information
- I repeated the test recording to a different stereo pair on the recorder and observed the same result repeated.
- I viewed the waveforms at the start of other bars and the same timing difference was observed.
- Both devices were returned to zero prior to recording start.
- Song position pointer was observed to work correctly from the hardware recorder to the Motif. ie. locate the hardware recorder to the start of any bar & the Motif sequencer also locates to the exact same position.
- The same timing discrepancy also occurs if I do a similar test in PATTERN mode on the Motif XF.
- For comparison, I did a recording with the same overall setup except this time the Motif XF was manually started and using its own internal clock. in this recording, the metronome audio and the sequenced drum audio were perfectly aligned to one another.
- The following images show my note data, playfx page, metronome settings and utility settings on the Motif XF.
Posted : 13/07/2020 5:07 am
Topic starter
Temporary workarounds to get a well-timed metronome "on tape" when the Motif XF is slaved to the Roland VS2000CD by MIDI clock:
All of these have been tested and verified to work.
- Make my own click track in the Motif XF sequencer and record that, OR
- Record the Motif XF metronome and then shift the audio track to get the metronome transients aligned with the bar lines on the recorder, OR
- Run a cable from the MIDI THRU port of the Motif XF to an input of my QY700 sequencer, set the QY700 as MIDI clock slave, turn on its metronome and record that (note: Roland multitrack is still the master device in this setup)
Posted : 14/07/2020 12:24 am
