Determining the best delay setting for external instrument

Hello,

I’m setting up two external instruments, inside Cubase 7.5, an Access Virus KC and a Korg Triton Rack.

I just tried with the Virus for now, and all seems OK, I just have to configure the delay setting for the external instrument, to compensate latency.
The problem is that I have difficulties in determining which is the best delay setting.

To see how much is the latency of the external instrument (Virus), I played some drum shots (“rimshot”) on the Virus, via MIDI written inside Cubase, and recorded the Virus at the same time. But because of MIDI jitter (I think), each recorded drum shot is slightly differently aligned, in respect to the MIDI track notes. Thus I don’t know which is the most accurately recorded shot.

Perhaps I could set the delay setting by ear, but I’m not able to judge differences of a few milliseconds, so I don’t know if 10.00 ms would be better than 12.00 ms (for example).

Do you have any suggestions for determining the best delay setting?

I was thinking of this: if the MIDI notes are randomly imprecise (between zero and max. MIDI jitter effect), as it seems the case to me, I could make an average of all alignments in a test recording (still with drum shots), and use that for the delay setting. This would have the advantage that the maximum difference between the recorded shot, and the MIDI note in Cubase, would become half of the maximum jitter latency; if the jitter latency would span (for example) between 0 and +4 ms, by making an average (2 ms) the actual jitter latency would span between -2 and +2, in respect to the MIDI note in Cubase (because by aligning to 2 ms, the audio data between 0 and 2 ms would be shifted before the actual MIDI note). This should make the MIDI jitter less noticeable or disturbing. Perhaps in a correctly synchronized track, MIDI jitter would be not noticeable, but I’m not sure about this. And anyway, the less, the better.

If instead the recorded drum shots are mostly accurate, and the MIDI jitter is minimal and happens only sometimes, then perhaps (not sure) it would be better to adjust the delay on the external instrument, so that the “left-most” recorded shot is aligned with the MIDI note in Cubase, and the “jittered” shots come next.
But this doesn’t seem the case to me; it seems to me that all recorded shots have a random jitter latency (from zero to a maximum which I’m currently not able to determine).

I think the first procedure (by calculating the average MIDI jitter) should work.
Or again, I should just try to configure the delay setting by ear?
Any suggestions? Other useful procedures?

Thanks.

At what buffer size is your soundcard set to ?

I have a MOTU 24I/O which was set to 2048 during the tests. But I won’t keep it always at this size (I want to try also buffers like 64 for low latency monitoring).

In my first post I did a little mistake.
I wrote: “To see how much is the latency of the external instrument (Virus), I played some drum shots (“rimshot”) on the Virus, via MIDI written inside Cubase, and recorded the Virus at the same time.”
In reality I did a real-time mixdown (which should be the same).