bvideo

It's possible to zoom all the way in to the sample resolution. See anything then?

IRQ conflicts used to be a thing. Now not so much. IRQs used to go to "15" on early PC hardware. Now, if you look at your "System Information" under the [-] hardware resources -> IRQs, you see numbers up to 511 at least. And no conflicts. This was taken care of by hardware improvements in PC architecture, including an "Advanced Programmable Interrupt Controller" and also "Message Signaled Interrupts". Software improvements include "Deferred Procedure Call" (DPC), which offloads traditional (old fashioned) interrupt processing to lower priorities (and also allowing for use of all the cores to do device servicing). Nowadays, when looking for hitches and sputters, it is frequently in some DPC, and Resplendence LatencyMon is likely to point out which ones cause the problems. Some older network interface drivers used to be somewhat of a problem, especially if your cable was unplugged, because they would run high priority loops to detect the cable status.

Ram makes a difference when you use a sample-playing vsti. It might be helpful to say a lot more about the projects you have: audio? midi? softsynths? number of tracks? sample rate?

How about "envelope segment shape"? You can change the straight lines between nodes to curves. Perhaps your envelope nodes happen at inconvenient points in the wavevorm where a sudden change in gain causes clicks. Perhaps using a curve makes the change not so clicky.

To eliminate certain artifact-producing settings, you could see if it happens with 64-bit audio set/not set and plug-in load balancing set/not set.

What counts more than anything is the input setting on every MIDI track. (User 905133's initial assumption).

What does your multidock look like? (That's where you expect the piano roll to open, right?)

Red-lighting the prochannel EQ is definitely worth avoiding when expecting it to show useful levels. But just for fun, I overloaded the master bus with an audio track & the master bus input gain. Reduced the master bus master fader until the master bus meter was never red. Then enabled the prochannel eq & set a mild curve. The prochannel overload light was full on. But I did not hear any artifacts while the audio played. So the red lights didn't seem to correspond with audible artifacts. Of course some other VST or prochannel effects could react badly when processing "out of range" data, even when processing in floating point, so there might be several reasons for not fixing master bus overload using the master fader.

In the analog mixing world, this would not be good. But in the digital - floating point - world, it would be fine. I believe the master bus output through its output fader is still floating point. Floating point numbers can represent much larger quantities than the fixed point 24 bit numbers that audio drivers work with. I assume after the master bus fader, floating point gets converted to 24-bit fixed point on the way to the driver. That's where clipping or terrible digital artifacts would first happen if the meter reads "too hot". Of course I don't know the code, and the signal path diagram in the manual is not necessarily a proof of this assumption, but it might be fun to "overload" a signal sent to the master bus and see how it sounds when compensated with the master fader.

Chasing for articulations needs a different implementation from regular tracks. In your case, there is no "note" outstanding in the articulation track, and by conventional chasing, no note needs to be triggered. But obviously articulation chasing would need to trigger the last articulation note (or maybe even the last several notes), even though the articulation note lengths have expired. A workaround for your posted case might be to extend all note lengths up to the beginning of the next note, but there is an obvious conflict with that approach also. Chasing of articulated notes needs to run in parallel with chasing of their articulations. I wonder if they have tried to implement it that way?

I doubt this mixer can be configured to split the signal the way you want. Typically, a smallish mixer has a separate send/receive path (e.g. "aux") that is nominally for patching in effects, but can also be used to send & receive from your computer while the mixer is used to mix all the audio into your sound system. This mixer doesn't have that. On the other hand, your Realtek onboard should have some way to take input from your mixer into the computer while also mixing it with computer audio out the Realtek into your sound system. That would provide the separation you need.

When you see "very high" what does your disk usage look like? Is your sample player is thrashing samples?

Someone I know killed his computer with compressed air. Static electricity? Overspeeding fans?

This is the exact symptom of how Cakewalk and Sonar have frequently misjudged the number of beats when promoting an audio clip to a loop, as rsinger said in the first reply.

My detailed reply is waiting to be approved. Bottom line, Korg M1 and Triton vstis are multitimbral. If the extreme is different, that's a shame. This word just in: the Korg Collection TRITON extreme owner's manual has the same words about setting the MIDI channel for each timbre.