How To Finally Solve MIDI Latency

[MIDI Life Crisis]

Good points. Even the inclusion of the pre-hit sound can confuse the ear. True Pianos (and I assume other VIs) let you decrease or eliminate that effect. Pre-delay in a reverb can also give the perception of a VI delay. There's a lot more to this than meets the ear.

With that, it is also important to mention that sometimes that natural latency that you describe is included in the samples. A careful sample editor will not trim any of those early attack sounds from the sample-- (i.e., before the string or air column is fully activated).

So if the sample includes some key/hammer activation sound, then when triggered, you have the actual MIDI key/hammer time interval, followed by the latency of the system, followed by the key/hammer noise in the sample, followed by the actual attack on the string.

All the more reason to try to eliminate that which can be eliminated, then acclimate to what is left.

[Shooshie]

What people don't realize sometimes is that every acoustic instrument has some amount of latency. It feels like the notes are instantaneous, and for humans they are. But it takes a micro section for a hammer to hit a string, a column of air to vibrate, or for a bow to get a sound going. Percussion probably has the shortest natural latency, which may account for so many inexperienced drummers having a tendency to speed up.

Not that this observation will help with computer latency, but the brain can adjust sufficiently to compensate for some latency. The thing is, once people realize there is ANY latency in a VI they (and I mean me, too) will notice it and try to make it zero. In fact, there will always be some latency in VIs and one needs to adjust one's brain to accept that. Yes, reduce it to acceptable levels, for sure, but it can never be eliminated, only reduced to acceptable levels.

This is interesting to observe. When recording solo piano, I noticed long ago that the attacks of the strings closer to the mics arrived sooner, and that you could predictably see the locations of the strings based on their arrival times. I used to record Yamaha Disklaviers almost daily, and their attacks are pretty much sample-accurate, otherwise this would be nearly impossible to observe. When you compare the MIDI note with the arrival of its sound, you can measure the speed of sound based on their arrival time in DP. (accurately, no less) This is assuming that the Disklavier is properly calibrated, and is professionally maintained.

Also, on the longer strings, the sound interferes with itself depending on where the mic is placed. I tend to place mics over the lyre where you get more of the ripple effect as the shock wave rolls down the string for the upper harmonics, as opposed to the slightly less interesting transverse waves that come with the mic placed at a tangent to the mid-point of the string.

There's a whole science of sound that gets opened up when you consider the latency within the instruments themselves, and it's one that contributes to the sound you record, so it ought to be something we think about when placing mics, players, gobos, and even studio treatments. I'm sure someone has written a book about how it affects drums!

Piano, btw, is a source of more tall tales and studio mythology than perhaps any other instrument, or at least 2nd to guitar. There are so many poor miccing positions that are perpetuated by engineers who use them because so-and-so said it's the best place to put mics on a piano. The Disklaviers offered me an objective laboratory for trying these out, which is admittedly a luxury not available to everyone. You can actually duplicate the same performance ad-infinitum to near-sample accuracy, so that eliminates the variable of the performance itself. What's left is simply mic placement. I recorded 6 mics at a time until I had 24 mic positions documented for various snippets of piano styles — lyrical, pop, technical, loud, etc. — then went back and compared them. Next I compared pairs, pairing pretty much every mic with every other mic (except where it just didn't make sense). The results were astounding. Many time-honored mic positions were rejected when compared with really good ones. For example, the mic under the sound board is a horrible place if you are recording solo piano. Maybe it would work for burying an ensemble piano in, say, a rock group, but wouldn't you rather have the whole sound and EQ it? Why limit yourself before you even mix?

Anyway, this is one of my favorite subjects, largely because of that research from 15 years ago. I still use the data.

Shooshie

[MIDI Life Crisis]

It is interesting. I have a few electric pianos and a Yamaha grand (G5). My Kurzweil PC2 does add a little bit of latency, but the Korg SP-280 has zero latency and they are very different pianos I use in very different settings. Of course the action is also totally different and that probably contributed to the difference as well, but the actual sample start times seem different, with the Kurz giving a slight delay, which feels more like a real piano to me.

Once I start playing, my brain adapts quickly to latency, as well as when I'm playing a VI. I can tolerate an awful lot of latency when tracking. It's a matter of precise timing, or maybe imprecise timing... LOL. Worst case scenario is having to move an overdub a millisecond or two after recording, but I haven't had to do that in a few years.

It may also be a matter of how bitchy the player is feeling that day.

[Timeline]

I just create a stereo mix and select 64 in another DP session but in the future I will try Shooshie's duplicate mix theory which sounds great. Less time maybe. Thanks S. for the idea reminder.

[daniel.sneed]

I set all my bus FX plugs (verbs, delays,...),in a V-rack and my master fader FX in another one (Ozone, mastering eq,...).
While tracking overdubs, I disable these V-racks and set buffer as low as 128, or 64 if required.
AFAICT, effective, reliable and done in a glance.

[Shooshie]

I set all my bus FX plugs (verbs, delays,...),in a V-rack and my master fader FX in another one (Ozone, mastering eq,...).
While tracking overdubs, I disable these V-racks and set buffer as low as 128, or 64 if required.
AFAICT, effective, reliable and done in a glance.

Yeah, just turning off that V-Rack can be very effective. Another reason I prefer to work with V-Racks. And just as an aside, you can make separate V-Racks for different kinds of instrumentation, and even for basic audio routing for any file. Though the auxes in a V-Rack won't have automation, you can drag a V-Rack into the Tracks Window, copying its tracks there as a quick way to set up an audio system.

Also, you can import V-Racks (as you can with any chunk) to any file with the Load function under the File menu.

Shooshie