Pro Tools LE™ Latency Charts
For Digi 002® and 002 Rack™
why Hardware Buffer settings of 256 or less
don't have to be a major problem
Pro Tools LE using a 002 interface has fairly significant tracking latencies unless you're using Low Latency Monitoring mode, which has its pros and cons, and is not always feasible.
It can be intimidating to grasp the relationships between sample rates, hardware buffer settings, and latencies measured in samples and in milliseconds. Documentation on the subject seems to be a little on the lean side. So I thought I'd run some tests and share the results. I hope that these numbers hold for all 002 users. It would be well to run a few tests yourself to make sure. Doing so will help you understand your system better. How I ran the tests.
|@ 44.1 KHz|
|LLM||87||1.973||2' 3"||68 cm|
|64||311||7.052||9' 0"||2.43 m|
|128||461||10.454||12' 10"||3.60 m|
|256||761||17.256||19' 6"||5.94 m|
|512||1361||30.862||35' 10"||10.62 m|
|1024||2561||58.073||66' 7"||19.98 m|
|@ 48 KHz|
|LLM||87||1.813||2' 1"||62 cm|
|64||306||6.375||7' 2"||2.19 m|
|128||458||9.542||11' 9"||3.28 m|
|256||762||15.875||18' 11"||5.46 m|
|512||1370||28.542||32' 3"||9.82 m|
|1024||2586||53.875||61' 11"||18.53 m|
|@ 88.2 KHz|
|LLM||88||0.998||1' 2"||34 cm|
|128||497||5.635||6' 4"||1.94 m|
|256||797||9.036||10' 3"||3.11 m|
|512||1397||15.839||18' 11"||5.45 m|
|1024||2598||29.456||33' 3"||10.13 m|
|2048||4997||56.655||64' 0"||19.49 m|
|@ 96 KHz|
|LLM||88||0.917||1' 0"||32 cm|
|128||497||5.177||6' 10"||1.78 m|
|256||801||8.344||9' 5"||2.87 m|
|512||1409||14.677||17' 7"||5.05 m|
|1024||2625||27.344||31' 11"||9.41 m|
|2048||5057||52.677||60' 6"||18.12 m|
Notice that while the number of samples latency varies a lot between sample rates, the latency in milliseconds, which is how we experience it, is very similar for a given buffer setting/sample rate. E.g., the shortest buffer setting for each sample rate produces a monitoring latency of about 1 or 2 ms, while the longest for each is in the 52-58 ms range.
Acoustic instruments have latency
While zero latency monitoring is ideal, moderate latencies don't have to be a big problem. Think about it this way: if you're jamming on acoustic guitar with someone who is sitting 4 1/2 feet from you, you're hearing him with a latency of about 4 ms. That's because - sound travels through air at about 1130 ft/sec. At 48 KHz sampling rate, a buffer setting of 128 produces less latency than 9 feet of air. When monitoring with headphones, the latencies produced by Hardware Buffer settings of 256 and less are comparable to those produced when monitoring through speakers in many realworld studio situations.
Of course, most of us are not used to hearing ourselves seeming so far away most of the time. It takes some getting used to. But it's not the end of the world, and while it helps to understand it, it's nothing to obsess over.
The situations below all assume a 44.1 KHz or 48 KHz sampling rate.
HWB=64: too small to matter much
If your HWB is 64, and you're monitoring through headphones, then the latency is the same as if you were monitoring a zero-latency mix through speakers and sitting with your ears 7 to 9 feet from the speaker cones. If you're monitoring through speakers, and sitting 3' away, HWB=64 increases latency to the equivalent of sitting 10-12' away. 7-12 feet is a pretty typical distance for musicians to sit from speakers in a control room. It's closer than they might be to each other on a large stage. It's much closer than the most distant members of a symphony orchestra are from each other.
HWB=128: perfectly workable
A HWB of 128 doubles headphone latency to the equivalent of sitting about 12 or 13 feet from speakers -- still not far from normal for a situation where three or four studio musicians are packed into a medium-sized control room. Monitoring through speakers at 5' with HWB=128, the effective latency is like being 17-18' from the speakers instead of 5' - about the distance some band mates on a large stage might be from each other. 12 feet or so works OK for most musicians, if they know to expect it. 18' is beginning to be a stretch if a tight groove is required. It can be overcome, but it's a difficulty, no way around it. You definitely want headphones at this latency.
HWB=256: difficult but not impossible
A HWB setting of 256, through headphones, is like sitting 19 or 20 feet from speakers, pretty far for a tight groove. Classical musicians should be used to such latencies, and jazz cats might be, but pop and rock musicians are going to need some time to get accustomed to that much delay. As long as they're hearing a good strong click that is not subject to the latency, they should be able to lock onto that with practice, and not be thrown too badly, very often. You definitely don't want anybody monitoring through speakers with this buffer setting if you can help it.
A HWB of 512 drives you up to the equivalent of sitting over 30 feet from speakers, which would make it difficult to stay in the pocket of almost any groove. It still wouldn't be a problem for someone playing glissed diamonds on a guitar or playing an instrument with a relatively soft attack, though. You don't want to track at 512 if you can help it unless the style or instrumentation is very timing-tolerant.
HWB=1024, of course, is extremely tough for tracking almost anything where instruments have to stay in time with each other if they're monitoring through PT.
The good news is, a very moderate power LE system can be tweaked to be able to lay down 18 tracks at once with HWB=128, as long as you lay off the plug-ins (which add more latency on armed tracks) until tracking the basics is complete. Bottom line, for good musicians, LE latencies don't have to be a serious problem if you have a good headphone cue system and understand how much latency to expect from a given buffer setting.
Pipe organs: bastions of latency
I once took organ lessons on a pipe organ in a large auditorium. The nearest rank of pipes was about 70 feet from the keyboard, and the farthest about 120 feet. The latency in the intervening air was about 60 - 110 ms, varying depending on which pipe was sounding. On top of that, the larger the pipe, the longer it takes for it to speak once its valve opens. For the lowest notes in the farthest rank, there was at least 300 ms of latency from the time the pedal was pushed until the player heard the sound.
Now this is an acoustic instrument in a real room, mind you, no digital nothin' nowhere. Compared to that, though, if 1024 HWB LE latency is the worst you've experienced, you don't know from latency. I was never very good at keeping all my pipe organ notes in sync; but my teacher, who played that organ almost every day, could rip out a Bach fugue at a tremendous tempo, and it was magic. How? He's a very good musician, he practiced his craft, and he adapted to the situation. That's what good musicians do. Instead of whining, I mean. :-)
DUC ID: daeron80
Written: April 2006
Copyright © 2006, all rights reserved
David's Personal Home Page