Old 07-20-2017, 08:03 AM   #1
karbomusic
Human being with feelings
 
karbomusic's Avatar
 
Join Date: May 2009
Posts: 24,354
Default Some intersting observations concerning latency

I've been working lately in such a way that concepts such as audio latency, groove, tempo, tightness, feel have come into focus. Mostly because I'm about to formally record my band which means we are making sure we are ready from a meter/tempo standpoint and so on - I have to basically transport my entire studio to the drum tracking location and well, we can't run into any roadblocks there (aka tempo/playing to click etc) since we only have 2 days to track drums for 12 songs. During this preparation, I've somewhat unintentionally stumbled across some interesting conclusions that address a few of the most repeated internet advice quotes paraphrased as...

1) "Once latency is less than 10ms, you can't hear the latency, which is backed up by science - therefore you are unnecessarily chasing a dragon that doesn't exist once your round trip DAW latency is less than ~10ms."

2) "It takes about 10ms for your eyes to translate what it sees, so visual cues (such as a band members watching each other) aren't going to help you sync up any better".

3) "Sound travels about 1 foot every .9ms, so even if you could hear sub 10ms latency, it wouldn't matter because you are already several feet from your monitors (DAW), or multiples of that distance in a band/stage setting".

Exhibit A



http://wallsonic.com/latency/ClicksL...ndAccuracy.mp3

This is a click, duplicated and separated by just a smidge over 2ms. The MP3 just under the image is a render of those tracks where one track is occasionally muted. It should be clear to anyone listening that we can hear the flams when both are playing. And when doing so it sort of sounds the same as a band who isn't quite tight as they should be, in time but not quite right. If everyone in the band were this far off (in various directions), it would still sound OK but kind of messy and not necessarily locked in - assuming what you are playing is assumed to be right on top of the beat and locked in vs playing behind/ahead purposely.

Though we are doing some hair splitting here, it's pretty clear we can hear and sense differences as small as ~2ms, possibly smaller. More interestingly, that musician who is "scary tight" can likely control timing down to as low as 1ms or less. Let's now revisit #s 1,2 & 3 above.

Quote:
"Once latency is less than 10ms, you can't hear the latency which is backed up by science - therefore you are unnecessarily chasing a dragon that doesn't exist once your round trip DAW latency is less than ~10ms."
That scientific truth is based on individual sounds. The first time I heard of it was back in the late 80s, concerning delay FX units where the difference between slapback delay and doubling delay was crossing that 10ms barrier, once we are under 10ms we can no longer hear two distinct sources but that does NOT mean the effect of those sub 10ms sources can't be heard as flams, not 'feeling' right or generally not as tight as it should be.

Quote:
"It takes about 10ms for your eyes to translate what it sees, so visual cues (such as a band members watching each other) aren't going to help you sync up any better".
Someone gave me a really hard time about this (in this forum) maybe a year ago. Turns out they were wrong. The science of the time it takes to process is correct however, since what you see is streaming at you in real time, you are able to anticipate and still lock in - for example being 10 feet away from the drummer while watching him.

I tested this recently by purchasing SoundBrenner Pulses, (they kick ass btw) syncing to a Galaxy tablet and strapping it on. I tend to place it on my shin bone so that it vibrates that bone. The reasoning is that when I tap my foot, as long as I'm "on it", I can't feel the vibration of the metronome at all and I can play as normal until I feel the vibration come back which means I've drifted off the pulse.

It is undeniably obvious when this occurs, I could readjust and get back to "dead on" and make the click disappear, however, I found two other interesting results...

A) if I watched the tablet which was about 7 feet away, I could use the visual feedback and stay locked in on more difficult passages. B) If I fell off the click, again, looking at the visual got me locked back in quicker every time as I had an additional sense involved. There were zero exceptions to this so I've concluded that visual feedback is doable, useful, credible and that any processing time the brain does is irrelevant. This shows that watching each other as needed in a band situation does make a difference - or watching the visual feedback in your DAW such as the transport clock may be quite useful.

Quote:
"Sound travels about 1 foot every .9ms so even if you could hear sub 10ms latency it wouldn't matter because you are already several feet from your monitors (DAW) or multiples of that distance in a band/stage setting".
Until DAWs and SIMs came around, much of the tracking was done with headphones. That means the monitor to ear latency is less than 1/10th of 1ms vs the average of 3 or more ms when listening to monitors - if it were pre-DAW we would also be missing its latency as well. If we take the latency we already have from the DAW, combine that with the 3+ ms coming from the monitors and consider the fact we can hear as little as 2ms difference in the music itself, all those 'crazy people' complaining about latency in the sub 5ms range, aren't that crazy after all. One thing I did test but didn't post here was the timing differences between tracking the same track twice, one through a SIM listening to monitors then another take with headphones (not mentioning since I didn't finish the test).

None of this means we can't make a successful recording, we can compensate in various ways before, during (even unintentionally) and after the fact. We can let the cards fall where they may and still be able to record a great piece of music. But... when there is a problem, if we understand some of this, we can address it, fix it or make it better. IOW, if I am wearing my creative/musical hat, not a damn bit of this matters to me unless it becomes a problem, but until someone shows otherwise, from the technical side of things, saying sub 10ms latency is irrelevant is sounding more and more like a myth to me - that doesn't make it a deal breaker either, just something good to know.

Discuss and/or point out anything I've missed. Just remember we are wearing our technical hats for a bit, not our creative ones.
__________________
If it requires a null test to find it, it is by definition minuscule.

Last edited by karbomusic; 07-20-2017 at 11:43 AM.
karbomusic is online now   Reply With Quote
Old 07-20-2017, 10:11 AM   #2
ashcat_lt
Human being with feelings
 
Join Date: Dec 2012
Posts: 4,133
Default

I think you will find that in context it's a non-issue. As long as your interface's drivers are honest with Reaper, it pretty much just works. Maybe I've never recorded anything that needed to be that tight and I definitely don't have a clue how they really do it behind the scenes. I have always just trusted that Reaper knows I'm hearing it late AND that it's hearing me late, and automagically somehow, when you play it back it's just like if you were tracking to tape. Sounds fine to me.

You can try a loopback test, but I'm not sure that's actually fair. Somehow it feels like there's an extra layer of something happening in there that makes it not exactly the same as how it works when overdubbing.

I am vaguely interested in learning more, but honestly it's always seemed to work fine.
ashcat_lt is online now   Reply With Quote
Old 07-20-2017, 11:16 AM   #3
karbomusic
Human being with feelings
 
karbomusic's Avatar
 
Join Date: May 2009
Posts: 24,354
Default

Quote:
Originally Posted by ashcat_lt View Post
I think you will find that in context it's a non-issue. As long as your interface's drivers are honest with Reaper, it pretty much just works. Maybe I've never recorded anything that needed to be that tight and I definitely don't have a clue how they really do it behind the scenes. I have always just trusted that Reaper knows I'm hearing it late AND that it's hearing me late, and automagically somehow, when you play it back it's just like if you were tracking to tape. Sounds fine to me.

You can try a loopback test, but I'm not sure that's actually fair. Somehow it feels like there's an extra layer of something happening in there that makes it not exactly the same as how it works when overdubbing.

I am vaguely interested in learning more, but honestly it's always seemed to work fine.
Yea, I'm well versed in the bulk of that, including loopback and how some users anticipate... some use what they hear and some use what they feel in their wrist to compensate - this is specifically important with automatic delay compensation because some players create that adjustment in their playing using different methods (which can counteract that compensation) and I've seen a number of threads where this came up but no one actually realized what the difference was, so it isn't always a non-issue.

Also, the often repeated no one can tell sub 10ms latencies simply isn't true (I've supported both sides of that argument but now I know the real answer), and understanding this in general would help those people. My main personal interest ended up being the finding that I never imagined really good players, are surely in that ~1 ms range of accuracy timing wise. Another point of interest that I'll gently present, if a persons timing resolution is 2ms or above, then none of this matters because they can't consistently perform any tighter that the latency itself.


*** Something I should clear up concerning the wrist statement et al. Some players strike their note, then listen for it to come back out of the monitors as their audial queue as to where the note lives in time and readjust their playing accordingly. However, others use the flick/vibration in their wrist as they create the note, which bypasses the monitoring part of the latency, thusly the note will fall differently on the timeline based on the method used.
__________________
If it requires a null test to find it, it is by definition minuscule.

Last edited by karbomusic; 07-20-2017 at 11:49 AM.
karbomusic is online now   Reply With Quote
Old 07-20-2017, 03:33 PM   #4
alanofoz
Human being with feelings
 
alanofoz's Avatar
 
Join Date: Sep 2009
Location: Oz - Blue Mountains NSW, formerly Geelong
Posts: 735
Default

The two clicks shown may be 2ms "apart", but they're 4ms "out of sync"! (If that means anything in this context...)
__________________
It's "its" except when it's "it is".

bunyipmusic.com
alanofoz is online now   Reply With Quote
Old 07-20-2017, 04:02 PM   #5
karbomusic
Human being with feelings
 
karbomusic's Avatar
 
Join Date: May 2009
Posts: 24,354
Default

Quote:
Originally Posted by alanofoz View Post
The two clicks shown may be 2ms "apart", but they're 4ms "out of sync"! (If that means anything in this context...)
Good catch. That could matter based on how it is considered - aka if that were a snare above, the second one was 4ms late. I'll chew on that for a bit.
__________________
If it requires a null test to find it, it is by definition minuscule.
karbomusic is online now   Reply With Quote
Old 07-21-2017, 02:41 AM   #6
Geoff Waddington
Human being with feelings
 
Geoff Waddington's Avatar
 
Join Date: Mar 2009
Location: Dartmouth, Nova Scotia
Posts: 3,286
Default

I've been fighting this battle for years

I grew up playing funk bass, complete with it's incredibly tight timing.

For practice I used to put my Franz metronome on 40, and clap my hands to make the click disappear by nailing it so tight.

Did that mofo exercise for 20 minutes a day, every day !!

The inside story goes that when they were deciding what timing resolution they needed for MIDI they asked the killer dudes of the day (Steve Gadd, Bernard Purdie, Harvey Mason, et. al.) to play a groove and then alter it the slightest amount they could, laid back and pushed.

The answer they got surprised them. Apparently these cats could consistently change the groove by 1 msec !

So, reasoning that most tunes were about 120, they decided to use 480 ticks/quarter, or, at 120 rpm, 960 ticks/second.

I've listened to all kinds of arguments, but it still comes down to the same thing (sorry rockers), it depends on your genre, funk, fusion, Reggae, Soca, Latin, etc. being the most demanding.

I finally got things right by getting a UAD Apollo, so I can put FX on my tracks while recording with sub 2 msec delay, close enough for me, YMMV
__________________
CSI - You can donate here: geoffwaddington.ca
Alpha software: https://stash.reaper.fm/v/36903/CSI%20alpha.zip
Reaper forum thread: https://forum.cockos.com/showthread.php?t=183143

Last edited by Geoff Waddington; 07-21-2017 at 05:47 AM.
Geoff Waddington is offline   Reply With Quote
Old 07-21-2017, 05:59 AM   #7
karbomusic
Human being with feelings
 
karbomusic's Avatar
 
Join Date: May 2009
Posts: 24,354
Default

Howdy Geoff!

Quote:
Originally Posted by Geoff Waddington View Post
The answer they got surprised them. Apparently these cats could consistently change the groove by 1 msec !

So, reasoning that most tunes were about 120, they decided to use 480 ticks/quarter, or, at 120 rpm, 960 ticks/second.
Yea that's interesting and makes sense. Thanks for chiming in.
__________________
If it requires a null test to find it, it is by definition minuscule.
karbomusic is online now   Reply With Quote
Old 07-29-2017, 10:56 AM   #8
Indirect
Human being with feelings
 
Join Date: Dec 2015
Posts: 120
Default

Try this. Offset bass part versus drum part for +/- 4 - 10 ms and see if groove changes. Sounds should be phase-unrelated.
Indirect is offline   Reply With Quote
Old 07-29-2017, 11:08 AM   #9
karbomusic
Human being with feelings
 
karbomusic's Avatar
 
Join Date: May 2009
Posts: 24,354
Default

Quote:
Originally Posted by Indirect View Post
Try this. Offset bass part versus drum part for +/- 4 - 10 ms and see if groove changes. Sounds should be phase-unrelated.
It's usually already there from the playing.
__________________
If it requires a null test to find it, it is by definition minuscule.
karbomusic is online now   Reply With Quote
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT -7. The time now is 03:50 PM.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2019, vBulletin Solutions Inc.