[mixaudio]

Hi MKZ, I've been in the exact same situation. I have done some testing too, trying to come up with my own delay chart, and there are so many variables, that I've decided just to do a quick "track to track" bouncing test for each particular session, when I need to check for a specific amount of delay.

What you're getting might have to do with wether or not your "low latency monitoring" function is engaged or disengaged. Then, when you test different multiple HW buffer sizes, and this will also affect MIDI data, once it becomes audio, specially for any external MIDI synths.

If you set a MIDI track to play data onto an internal Virtual Instrument, you won't get an important amount of latency, since PT does some automatic compensation (PT does not do ADC for plug-ins, but it actualy does for MIDI running into VIs). Then, if you measure the amount of audio delay you get from your external MIDI modules or keyboards, when they are being triggered from MIDI tracks in ProTools, you'll notice that the audio is coming back to PT with a certain amount of delay that is prety much the amount of samples you have set in the HW buffer size. So, say for instance, that you have MIDI track for a steady, kick drum going to an external module, and your HW buffer size is set to the highest level (1024), and then you print the audio of this external MIDI kick drum onto a PT audio track, you'll be looking at a waveform that appears about 1400-1500 samples after the actual MIDI hits. This is an issue I've been hoping that will be solved soon, but haven't heard any specific comment from DigiTechSupp, and the only way to solve it is to set individual negative amounts of delay for MIDI tracks running to external MIDI synths (and leave MIDI tracks for internal VIs set to 0) either from the MIDI Realtime Properties window, or from the global MIDI delay window. Unfortunately, there is no "Realtime Properties" for audio tracks that will let you specify a negative amount of playback delay for audio regions (I would love that, even if it was not "automatic").

Then, you also have the variable of sample rate. You won't necessarily get the same delay amounts when working at 44.1K or 48K.

Also, if you need to print several MIDI tracks that run to independent external synths, but at the same time you need them submixed onto a single audio track, and you are not using an external mixer, then you'd have to hook them to PT analog inputs, and run them thru several Aux inputs which would be bussed digitally onto the destination audio track. Well, by doing this, you'll get a double amount of delay, because first, you'll have the MIDI data going to those synths being sent late (by the amount of HW buffer size), plus the latency created by bussing the Aux tracks to send the signal internally to the audio track where you'll be printing. So... with a HW buffer size of 1024, you'd be looking at your MIDI data becoming printed as audio, at about 2000 to 2500 samples late.

It's confusing... but the worse is that we are at the mercy of a glorious PT Engineering-Development Team to get a fix for this. Excuse the sarcasm, I actualy respect and love their work, but the reality is that we can not tell for sure of whether or not this is a priority for them or for Digidesign.

My couple of "ranty" 2 cents.

Rodolfo F.