If the MIDI controller is connected via USB and is polled at 1 ms, you have only about that amount of lag. This is not something most people can even notice.
Also, consider that even playing an acoustic piano has quite some lag: your brain has to send a signal to the muscles in your hand, which have to contract, which presses the key which takes some time to travel down, the hammers needs some time to travel to the strings, then the sound takes some time to travel to your ear. All this adds up to many milliseconds of delay. A musician already compensates for all that by starting the hand motions earlier to compensate for that lag. As long as it is not too large and if it is consistent, a few milliseconds extra does not matter.
So 3 milliseconds is about how long it takes sound to travel 1 meter.
So it's like the difference between how far a pianist sits from the strings in an upright vs a grand piano. Or the guitar player taking a couple of steps away from their amp. Or the difference between a drummer hitting the snare or their crash cymbal.
One of my local theatres says the stage is 19m wide, so an orchestra playing there is probably 50 ms apart for onmstruments playing on opposite sides of the stage.
Hell, you've got most of a millisecond of "lag" between your left and right ears for sounds coming from the side.
Submillisecond differences for audio > 1 KHz tend to show up as phase differences (after all: 1 ms period == 1 KHz) which allow you to figure out where something is located.
The difference between your right and your left ear is indeed on that order and so phase change really starts at roughly that frequency and below that it all just blends together. That's why bass speakers aren't directional and you can have a single big woofer in the middle and it won't make much of a difference in what the sound is like in the stereo image. But if you collapse your high or mid then suddenly all sense of position is gone.
So anything below 1 ms is negligible, as long as it isn't so long that it starts to confuse you. Not rare in church organs, where the lag can be so long that you wonder if you've depressed the key at all, you get used to it after a while but it is a very weird experience.
It's interesting that you can spot an out-of-phase wired woofer fairly easily by just moving around the room, if the sound of the bases cancels out in the middle that usually a good sign that either woofer has its polarity reversed from what it should be. The sound will be 'hollow' (I have no better term for it, that's what it sounds like to me).
My suspicion is that nobody could tell the difference between sub milliseconds, and several or even tens of milliseconds so long as the latency in both channels is the same. The stereo imaging isn't going 5to change if the L/R signal transient is at 0.000/0.0001ms, compared to 3.000/3.0001ms. The phase difference is the same.
(Surely that church organ delay is an artefact of the pneumatics, not the speed of sound travel time between the player and the various pipes?)
that Church organ delay is very much an artefact of the speed of sound travel time: the pipes sound half the cathedral away from the organist. In some churches this is made even worse because there is a conduit across the top of the vaulted space injecting sound into the other end directly, so then you get both a direct wave and a diminished one with a massive delay.
Super confusing, the best way to deal with it is to just put on headphones and only listen to the direct sound.
Ah interesting, thank you. I know about the missing fundamental because I run into that when tuning the low end of the piano and when trying to re-create the lowest notes on the piano from audio files. But I never made the connection to the canceled out bass notes but it is obviously the exact same thing.
Another metric to think about if you care about rhythm. In a 4/4 beat at 120 BPM with 50% swing (i.e. straight 16th notes with no swing), sixteenth notes are 125ms apart.
Each percent of swing is 2.5ms. So if you increase to 51% swing, it's 127.5ms and 122.5ms between successive pairs of 26th notes. 55% swing (which is very distinct from 50% to my ears) is 137.5ms and 112.5ms.
Also, consider that even playing an acoustic piano has quite some lag: your brain has to send a signal to the muscles in your hand, which have to contract, which presses the key which takes some time to travel down, the hammers needs some time to travel to the strings, then the sound takes some time to travel to your ear. All this adds up to many milliseconds of delay. A musician already compensates for all that by starting the hand motions earlier to compensate for that lag. As long as it is not too large and if it is consistent, a few milliseconds extra does not matter.