Random Midi Notes missing

[Guillermo Diaz]

Hi, anyone have this issue?

I have a large midi project, and i have midi notes dropouts, even if i mix the song, this dropouts are random, things that never occured preivously.

I hava a midi buffer size in 1000 ms, i read another discussion, and apparently, then can increase more, but in my computer dont let me pass the 1000 ms barrier.

Help me please, i have work to do.

I have a Icore i5, 16gb ram, Win10 64bits.

 

Thanks!

[David Baay]

MIDI buffering was revised a couple years ago, and those extremely high buffer settings are not generally needed. i think the default now is 50ms. But it'd still sensitive to plugins that have very large 'look-ahead' buffers.

If you're aware of any 'mastering' type plugins in your project that have big internal buffers, you might try disabling them. Some have 'quality' or 'precision' settings like Cakewalk's Multiband Compressor that can be lowered to reduce the buffer size.

[Guillermo Diaz]

 

12 minutes ago, David Baay said:

MIDI buffering was revised a couple years ago, and those extremely high buffer settings are not generally needed. i think the default now is 50ms. But it'd still sensitive to plugins that have very large 'look-ahead' buffers.

If you're aware of any 'mastering' type plugins in your project that have big internal buffers, you might try disabling them. Some have 'quality' or 'precision' settings like Cakewalk's Multiband Compressor that can be lowered to reduce the buffer size.

How can i check this option? i have Ik multimedia White 2A in some fx, everything are simple eq 

[David Baay]

White 2A is a Linear Phase EQ which means it will be using a lookahead buffer. Clicking the 'power' button on the FX in the bin to disable it should be enough to check whether it has any bearing. just make sure you get all of them. The effect is not additive, but leaving even on active will drive the delay compensation which is what causes trouble with MIDI buffering.

[Guillermo Diaz]

White 2a is a compressor, but I disable all the fx and still have the issue 

I already email to cakewalk hoping they can help me 

[David Baay]

Ah, OK. I misunderstood "White 2A in some fx, everything are simple eq", and my search hit on IK's LP EQ.

In any case. I would guess the issue is related to a specific plugin(s) whether FX or Instruments. Just having a lot of MIDI tracks should not drive up the buffering requirement.

Any MIDI FX in the project?

[Guillermo Diaz]

Oh im sorry for not to explain well.

No, i dont have any midi fx, I use Sampletank 3 and migrate all the sounds and midi channels to sampletank 4 and seems to free some ram, but the issue persist.

[David Baay]

If you were already at that high MIDI Prepare Using  buffer setting, you might actually want to try backing it down to the default of 50ms, and going up gradually from there.

Another possibility would be that tempo and note position are interacting in a way that causes buffering to fail. I have not seen this since buffering was revised, and usually only with one particular note in a whole project, but it might be worth tweaking the position of a dropped note by a tick or two and seeing if it makes a difference or tweaking the nearest tempo value before a missed note (or for the the whole project if it's a fixed tempo.).

I may not have the samples you are using, but I have Sampletank 4, and could give your a project spin on my system to see if it's platform-specific, and poke around for other possible causes if the problem reproduces.

Bakers will probably need to do the same to understand what's going on.

P.S. In thinking about this, I was reminded of an issue I encountered when I first got Sampletank that some instruments were not playing notes above a certain velocity because the channel-mapping of velocity layers in some patch definitions was screwed up. IIRC, IK responded to my post about it, and were going to fix them. Might have been specific to Syntronik patches though. The plain text patch files could be edited to fix the problem (putting all layers on channel 1), but there were too many so I only fixed the one I was using at the time.