Latency test of 23 amp modelers

jrockbridge

Member
Messages
5,294
Quick update re: my earlier post about guitar => wireless => modeler => active monitor => air => ears. Without the wireless, no issues. With wireless, major distraction. I was maybe 10 feet away from the monitor. The modeler's latency was ca. 2.8ms, monitor .65ms, TOF ca. 8.8ms. Total latency w/o wireless - approximately 12.25ms.

I just measured the latency of my wireless; it's 5.0ms, and adding it definitely took me across a perceptual threshold. Ergo, I'm OK with ca. 12ms latency, definitely not OK with 17ms.
This aligns with my experience. 12ms of latency was not a problem for me.

You may try another experiment. Keep playing with the 17ms latency. See if you can get used to it if you keep playing for awhile. Time yourself to see how long it takes for you to adjust and get used to 17ms of latency. You may feel like you never get used to it.

In my past experience, playing with 20ms of latency got easier the longer I continued to play. But, each time I sat down with my DAW with 20ms of round trip latency, I would feel disconnected and I’d have to play for awhile to get back in the swing of it. I felt like the longer I played, the less it bothered me.
 
Last edited:

Digital Igloo

Member
Messages
5,065
It should be more like kombucha instead of moonshine, cus signal processing latency with a decent digital gear setup tends to be smaller than TOF. So yeah, beer’s the one that’s mostly responsible for crossing the perceptual thresho… for passing out.
It's still moonshine, because your brain can't account for it. Kombucha might indeed make you vomit, but that's because it tastes like sour, fermented feet.



For most people, any disconnect from a latency of... let's go with Jay's 17.25ms, the equivalent of standing 19.5 feet away from a speaker... stems from the fact that one's brain has difficulty resolving that the speaker isn't REALLY 19.5 feet away.

For the sake of our collective livers, let's cut the alcohol in half, so 0.5 drinks per millisecond:
  • Jay's guitar > wireless > modeler > active monitor > 10 feet of air = 17.25ms (4.2 shots of moonshine + 4.4 beers)
  • Guitar > amp > 19.5 feet of air = 17.25ms (8.6 beers)
The same person could stand 19.5 feet away from their tube amp with a long guitar cable and get the exact same 17.25ms of latency, except they'd more than likely play just fine. They'll instinctually know what standing 19.5 feet from their amp should sound and feel like. Inexperienced guitarists' palm mutes might suffer a bit within the context of a band, but they shouldn't feel any disconnect and certainly wouldn't dream of complaining to the companies who made their guitar and amp.

Or go the other direction. Say one has Jay's rig above and instead of playing 10 feet away, they walk right up to the speaker and place their ear 1 foot away. Their accumulated latency is now down to 9.4ms, but chances are any latency disconnect won't magically disappear because their brain is now resolving for something that appears to be right in front of them but it sounds and feels like it's 10.6 feet away. Even if they're not sure exactly what's going on, something's... off.

The difference is how the accumulated latency is achieved—one is natural and wholly expected by the human brain for tens of thousands of years; the other is unnatural, unexpected, and sometimes jarring. That's why low latency in multieffects and DAW interfaces is important. It's not THE most important thing, but it's still a big deal.

It's not a multieffect, but here's a big reason why I chose Lynx Aurora (n) for the new studio:
Interface Latency Table.jpg
 
Last edited:

Jay Mitchell

Member
Messages
5,827
You may try another experiment. Keep playing with the 17ms latency. See if you can get used to it if you keep playing for awhile.
I have no doubt that I could, but there's no need for me to do that. Since the whole problem was brought on by the addition of a piece of gear I normally wouldn't use - the wireless - I never bothered getting used to it. The only reason I brought this up again was that I still have the wireless system lying around, and I was interested in seeing how much latency it added.
 

Jay Mitchell

Member
Messages
5,827
For most people, any disconnect from a latency of... let's go with Jay's 17.25ms, the equivalent of standing 19.5 feet away from a speaker... stems from the fact that one's brain has difficulty resolving that the speaker isn't REALLY 19.5 feet away.
That's a highly speculative statement, and my money says it's mistaken.

In my case, the distraction came from the fact that my tactile feel for the attack when I picked/strummed didn't match what I was hearing. IOW, it was a timing discrepancy, albeit a relatively subtle one. I'm sure I could get used to that much latency, but why bother? I could probably get used to playing 20 feet away from my cab, too.

FYI, in most scenarios that place a guitar player 20 feet or more from his/her cab, there will be a wireless system and one or more wedges that are much closer to the player. The player will get most of their volume from the wedges. Even with the 5ms from the wireless - I'm sure the better ones have less than that - the total latency in the wedge signal is substantially less than TOF.

The rest of your post is psychoacoustic speculation that I believe to be mistaken.
 

Digital Igloo

Member
Messages
5,065
The rest of your post is psychoacoustic speculation that I believe to be mistaken.
It's easy to test if you have a traditional amp and a 20-foot guitar cable. Is playing from 19.5 feet away equally jarring for the same reasons or is the experience what you might expect? Or is it just different?

The other experiment would be to walk up to the speaker, put your ear 1 foot away and see if the disconnect you felt from 10 feet away disappears completely.

Psychoacoustics is a fascinating subject. My wife's cousin is an Acoustic Engineer at Apple and the conversations are wild. No, he won't leak any product info.
 

PLysander

Member
Messages
532
It was your example - it isn’t my fault that you chose an analogy that is incapable of describing latency sensing issues.

Dude. It was your example, which presented relative phase as a gotcha:

This is analogy actually is trying to press the cube through the round hole.

You're right that one might not easily hear a difference between 100 and 101 bpm. You might hear a difference of two streams running more and more out of phase of course... quite quickly actually.

Once again, and to be perfectly clear, this is 100% true... and would have nothing to do with latency. At all. Zero. Nada. This is not my opinion - is just the way physics work.

This is still a complete straw man. Nobody argues that one would hear 1ms differences in metronomes.

If you can't hear a beat being a couple milliseconds late in your metronome, then you cannot hear your instrument's sound being delayed by the same amount. That's the entire point of the exercise.

You might think you can still hear it - and that's perfectly fine. But those are not your ears at work.

Reaction time is not the same as the time correlating different sensoric impressions. You again only talk about sound. Playing an instrument is more than hearing something. You might try playing your guitar with thick padded winter gloves. If you still play perfectly well and perceive absolutely no difference to playing without gloves - then we can talk again ^^

Oh yes, it would likely feel like crap, but i would also make no difference regarding how long the (now crap) sound it creates takes to reach my ears. We're discussing latency, and how it impacts how you listen to what you play.

I need to quit this thread; i feel like everyone's talking in circles by now.
 
Last edited:

jrockbridge

Member
Messages
5,294
It's still moonshine, because your brain can't account for it. Kombucha might indeed make you vomit, but that's because it tastes like sour, fermented feet.



For most people, any disconnect from a latency of... let's go with Jay's 17.25ms, the equivalent of standing 19.5 feet away from a speaker... stems from the fact that one's brain has difficulty resolving that the speaker isn't REALLY 19.5 feet away.

For the sake of our collective livers, let's cut the alcohol in half, so 0.5 drinks per millisecond:
  • Jay's guitar > wireless > modeler > active monitor > 10 feet of air = 17.25ms (4.2 shots of moonshine + 4.4 beers)
  • Guitar > amp > 19.5 feet of air = 17.25ms (8.6 beers)
The same person could stand 19.5 feet away from their tube amp with a long guitar cable and get the exact same 17.25ms of latency, except they'd more than likely play just fine. They'll instinctually know what standing 19.5 feet from their amp should sound and feel like. Inexperienced guitarists' palm mutes might suffer a bit within the context of a band, but they shouldn't feel any disconnect and certainly wouldn't dream of complaining to the companies who made their guitar and amp.

Or go the other direction. Say one has Jay's rig above and instead of playing 10 feet away, they walk right up to the speaker and place their ear 1 foot away. Their accumulated latency is now down to 9.4ms, but chances are any latency disconnect won't magically disappear because their brain is now resolving for something that appears to be right in front of them but it sounds and feels like it's 10.6 feet away. Even if they're not sure exactly what's going on, something's... off.

The difference is how the accumulated latency is achieved—one is natural and wholly expected by the human brain for tens of thousands of years; the other is unnatural, unexpected, and sometimes jarring. That's why low latency in multieffects and DAW interfaces is important. It's not THE most important thing, but it's still a big deal.

It's not a multieffect, but here's a big reason why I chose Lynx Aurora (n) for the new studio:
View attachment 644036
I’m glad you posted this chart. I happen to be putting a PC DAW together right now. I wish they had done the chart differently because this shows a ranking number rather than a latency number. If the difference between the top ranked and the bottom is only 1.5 ms in latency, it becomes less critical to me. But, I guess they did not want to use latency numbers since they vary on each interface depending on other details of use.

Focusrite is on the bottom of the chart. They are a popular consumer grade interface for a lot of people on TGP. I’m sure their low prices play a big part in the popularity. I went to their website and they dedicate a lot of text to discussing round trip latency. The latency with an audio interface depends on a lot of factors including the computer, the operating system, the multitrack software being used, the buffer size, etc. A quick glance at their chart with various setups seemed to indicate that round trip latency in most cases would anywhere from less than 3ms to over 9ms. So, it would seem that even a low ranking, current model, focusrite audio interface would have round trip latency that’s pretty low. But, of course, as soon as you add something else to the signal path, a modeler, a wireless, a plug-in, it could add some latency. So, if a top of the line audio interface can shave 2-3ms of latency over a low ranking interface, it could make a significant difference in the final feel.

I am definitely starting to understand why picking low latency devices could end up being significant. Each device can add a few more ms of latency to the overall chain. With a DAW, the round trip latency could be the difference between being able to nail an overdub or not. Personally, as a hobbyist, I just want to be able to do a half way decent job of laying down some basic tracks. I’m not trying to get a professional result.
 

aptfx

Member
Messages
245
Dude. It was your example, which presented relative phase as a gotcha:

You mix things up. It was your off topic example and I presented that in _your_ off topic example relative phase is a gotcha. The point was not that it is, but that your example is absolutely off topic and unsuitable on the topic of latency.

Once again, and to be perfectly clear, this is 100% true... and would have nothing to do with latency. At all. Zero. Nada. This is not my opinion - is just the way physics work.

Again again - You may repeat it and make it look as being right. I do not doubt the physics (I brought it up) - the point is that YOUR example in incapable as an example about latency. So you're trying to "off topic" me with your own "off topic" analogy which is kinda funny.

If you can't hear a beat being a couple milliseconds late in your metronome, then you cannot hear your instrument's sound being delayed by the same amount. That's the entire point of the exercise.

You might think you can still hear it - and that's perfectly fine. But those are not your ears at work.

*sigh*. Hear. H.e.a.r. You always talk about "HEARing". I've already described this multiple times now. Latency is not only about hearing. You just have this idea that the whole point here is the time sound travels through air (and processing) and how this time is small enough to not _hear_ a distinctive difference between discrete events short enough separated. But the actual point here is that relational time domain calibration in your brain between sensoric events in your e.g. hand and audible events through your ears works differently well in different situations. This is not about "hearing a difference". Its your brain being ****ed up a bit getting this two sensoric events together and making sense of it. This is where psychoacoustic effects play a huge role. Its not about "imagining things that are not here" or "fake perception of something". The sensoric impression is damn real - its just that you brain processes it in a way can be misleading because the sensoric impression doesn't fit what your brain expects.

And yes I definitetly _sense_ (realize that there is something different) a difference in latency if playing through e.g. headphones with a round trip delay of e.g. 3ms vs. 8ms. No this does not mean that I "hear" a difference between those two settings. But the complete sensoric impression while playing is different. Its actually even quite easy to debunk your idea that this is just my brain making things up: I happened several times to me that the settings of my audio interface changed the buffer size from what I've set it... with essentially a difference of 3ms RTL to 8ms RTL. And in any of those cases I realized it quite quickly when playing that something is off. Why would I even get an idea that there is something wrong if it all is just made up from my brain? There was no reason to believe that there is something wrong - particularily the first time this happened.
 
Last edited:

PLysander

Member
Messages
532
(...)
*sigh*. Hear. H.e.a.r. You always talk about "HEARing". I've already described this multiple times now. Latency is not only about hearing. You just have this idea that the whole point here is the time sound travels through air (and processing) and how this time is small enough to not _hear_ a distinctive difference between discrete events short enough separated. But the actual point here is that relational time domain calibration in your brain between sensoric events in your e.g. hand and audible events through your ears works differently well in different situations.
(...)

I don't know what to tell you man. Unless you're referring to "sensoric events" that happen while you're connected to the Matrix, the point still stands.

I'm out.
 

Digital Igloo

Member
Messages
5,065
I’m glad you posted this chart. I happen to be putting a PC DAW together right now. I wish they had done the chart differently because this shows a ranking number rather than a latency number. If the difference between the top ranked and the bottom is only 1.5 ms in latency, it becomes less critical to me. But, I guess they did not want to use latency numbers since they vary on each interface depending on other details of use.

I am definitely starting to understand why picking low latency devices could end up being significant. Each device can add a few more ms of latency to the overall chain. With a DAW, the round trip latency could be the difference between being able to nail an overdub or not. Personally, as a hobbyist, I just want to be able to do a half way decent job of laying down some basic tracks. I’m not trying to get a professional result.
Yeah, that chart is pretty old; got it from Gearspace a while back. I'm sure the numbers are very different now, although RME always seems to top the charts. Unfortunately, they don't make anything with the I/O I need.

The Aurora (n) has been out for several years too, but it's rock solid, the drivers are impeccable, everyone I've spoken with adores the way it sounds, and Lynx customer service, even in presales, has been superb. Mine won't ship for another few weeks at the earliest. Not a big deal, as the new Mac and monitor were 11 weeks (!!!) out when ordered in mid-May.

Low-latency interfaces may not be important if you can deal with switching back and forth between your DAW and the control panel (if you're not using PT or Luna) and can put, say, Logic or whatever into low-latency monitoring mode (where severe latency-inducing plugins are bypassed), but if you record other people and don't want to slow down a session, they're pretty much critical.
 
Last edited:




Trending Topics

Top Bottom