Perceivable Latencies

[analographi]

What latencies are perceivable by human beings.

Do you know any ressources of knowledge about this?


What latency has a piano or a hammond?

How about a guitarist who stands 3 meters away from his amp, that involves 9ms of latency. Does he know that?

[matt thomas]

Originally Posted by analographi

What latencies are perceivable by human beings.

Do you know any ressources of knowledge about this?


What latency has a piano or a hammond?

How about a guitarist who stands 3 meters away from his amp, that involves 9ms of latency. Does he know that?

ha, interesting questions, I'm going to make up the answer as I go along, but hopefully it'll offer something to think about

latency is perceivable in different ways, so their are different answers.

The human ear can hear two distinct sounds about 30 ms apart, closer than this they just sound like the same sound. Now this isn't latency, but I would imagine a similar theory applies. (edit: well I thought it was that far, apparently I'm wrong according to the quote below)

However at times less than this you can hear the delay as the wave comb filtering if combined with a non delayed version of the sound. (think a flanger) This type of latency is perceivable almost right down to 0.

The flam between two instruments is like this as well. You hear a tonal difference as opposed to a timing difference when it is small.

You asked what is the latency of a piano for example.

In one way of looking at it the latency is zero, the sound starts exactly when its meant to. But there is more to it. Depending on how the action of the piano is designed the notes will actually sound at a different part of the keys travel. The sound never happens immediately when your finger touches the key, and I don't even think its when it as the bottom of its travel (surely there is at least some overshoot) it is always part way through its travel. For this reason we are used to hearing the sound only roughly when when expect it. And it follows that short latencies will not be noticeable when using a digital piano.

Furthermore like you said, distance from a sound source creates latency, we don't really perceive this delay at varying close distances, therefore we won't percieve small latency changes either (at least in the form of delay)

food for thought
narco

edit ps. I just found this after a google, (Nick Herbert, Elemental Mind, Dutton, 1993, p. 50.):

How finely can we divide our little 3-second lives? The shortest perceivable time division – sensory psychologists call it the fusion threshold – is between 2 and 30 milliseconds (ms) depending on sensory modality. Two sounds seem to fuse into one acoustic sensation if they are separated by less than 2 to 5 milliseconds. Two successive touches merge if they occur within about 10 milliseconds of one another, while flashes of light blur together if they are separated by less than about 20 to 30 milliseconds.

[The Byre]

Look up the Haas effect!

[IzzyRock]

Its funny, but you sure can hear 9 ms of latency in your DAW, how come its not noticeable in real life?
Interesting topic

[Bassmankr]

There are already threads on latency but when playing keyboards I really notice it when latency is over 10ms, keep it at 6ms or less to keep the players happy when tracking. You are comparing apples and oranges though when you are talking about the ear hearing two signals spaced Xms apart versus pressing a key or picking a string or singing into a mic and then hearing the sound it produces. Any delays with the later are VERY noticable to the talent and why people prefer analog monitering or a digital chain that basically goes from input to direct output skipping the rest of the digital signal path. Players will ajust to a small amount of latency but after a certain point it's useless.

[matt thomas]

I just found a "sound on sound" article on latency, the guy does some tests of various setups

However, playing a soft synth in real time can never offer sample-accurate timing, and my initial measurements show that you can expect anywhere between 3ms and 15ms to be added onto the reported soundcard latency before you hear a note, simply due to the MIDI interface and Windows operating system. This rather explains why some musicians claim to find soft synths sluggish even when their soundcard buffer is set to a typical 10ms -- the truth is that this setting may well result in a 25ms latency between playing a note and hearing the result, which falls into the 'noticeable' area.

Most musicians can easily adjust to latencies even as high as 15ms, as long as they are reasonably consistent -- it's the jitter that tends to be more problematic, as this determines the amount of 'looseness'.

from: THE TRUTH ABOUT LATENCY

narco

[drBill]

??

All I know is that I took out a drummer who's building a studio to a PTLE 003 studio yesterday. We were both concerned about latency, monitoring THRU the converters and how that might/would affect the groove of the rhythm section.

We started at a high buffer and worked our way down. When we got to 64 samples he thought it "might" be OK. When we got to "32" he said it was OK. Then I played him "input" (??) or whatever you want to call it - monitored straight thru the console - no conversion. That was the end of the conversation. Now he needs a HD system instead of a native one. Oh well....it's only money. For some, the latency of a native system - at ANY amount is too much.

[matt thomas]

Originally Posted by drBill

??

All I know is that I took out a drummer who's building a studio to a PTLE 003 studio yesterday. We were both concerned about latency, monitoring THRU the converters and how that might/would affect the groove of the rhythm section.

We started at a high buffer and worked our way down. When we got to 64 samples he thought it "might" be OK. When we got to "32" he said it was OK. Then I played him "input" (??) or whatever you want to call it - monitored straight thru the console - no conversion. That was the end of the conversation. Now he needs a HD system instead of a native one. Oh well....it's only money. For some, the latency of a native system - at ANY amount is too much.

at those latencies the daw will actually be MORE in time than listening to the drums acoustically from the drummers perspective.

for example. The distance between the snare and drummers ears is approximately 50 cm (probably more). It will take the sound of the snare drum 0.0015 seconds to reach the drummers ear (at 330 m/s)

The mic on the other hand will be about 10 cm from the snare drum (0.0003 seconds). The 32 sample latency at 44.1k will add 0.00007 seconds, the distance from his headphones to his ear will be about 2cm (0.00006 seconds). The electrical latency is negligible. giving us a grand total of approx 0.0004 seconds

So the monitoring done through 32 sample soundcard is 0.0004 seconds as opposed to 0.0015 seconds in real life, the DAW is 4 times more accurate

just out of interest

narco

(ps I didn't double check my maths, but I'm sure its at least in the right ball park)

[synthoid]

I'm a keyboard player, and I've learned over time to play soft keyboard sounds that have huge latencies, so long as the jitter is very low. I've been really surprised at what apparently happened inside my brain, haha, as this happened. I've gotten so that I don't realize that there is a latency -- my hands compensate for it automatically, and I can align my playing with other tracks with good accuracy -- the same as when there is no latency.

But the really weird thing is this. If I play sounds with long latency for a while -- say an hour -- and then I switch to a keyboard with zero latency and record with it, I have the sensation that the sounds are coming out of the keyboard before I play the notes! haha. I have several times stopped a take because I think that there is a track already printed with what I'm playing, because of this effect. Very weird.

Anyone else notice this?

-synthoid

[drBill]

Originally Posted by narco

at those latencies the daw will actually be MORE in time than listening to the drums acoustically from the drummers perspective.

for example. The distance between the snare and drummers ears is approximately 50 cm (probably more). It will take the sound of the snare drum 0.0015 seconds to reach the drummers ear (at 330 m/s)

The mic on the other hand will be about 10 cm from the snare drum (0.0003 seconds). The 32 sample latency at 44.1k will add 0.00007 seconds, the distance from his headphones to his ear will be about 2cm (0.00006 seconds). The electrical latency is negligible. giving us a grand total of approx 0.0004 seconds

So the monitoring done through 32 sample soundcard is 0.0004 seconds as opposed to 0.0015 seconds in real life, the DAW is 4 times more accurate

just out of interest

narco

(ps I didn't double check my maths, but I'm sure its at least in the right ball park)

Math, schmath. Say what you will, there was no fooling him. Once he played with zero latency, he could always feel when I switched over to monitoring thru the DAW. The brain is a funny thing.

Once I did an album with a very well known guitarist who shall remaiin nameless. He asked me to move a bass part 40 samples. I'm thinking, WTF, there's no way you can hear 40 samples, so I did what any self respecting engineer would do, I said I moved it before I did. He could tell and busted me on the spot. I did some blind A/B tests for him. 9 times out of 10 he could tell. At that time I absolutely couldn't hear the difference. But I'll tell you what, after spending a month working on a song that had some of the best of the best players on it, with this guy producing, I got to where I was feeling the groove differences between 20-50 samples - depending on the part and whether it was percussive or not. Now, without pulling your mind into a pretzel every time you analyze groove, etc., I don't think I could go there again without some serious hardcore painful listening. But it IS doable, and for him, it was necessary to find his "pocket" for each player in the band. You can probably imagine just about how fun those sessions were....

The mind is an amazing thing, capable of far more than we give it credit for. I have no doubts that this drummer was feeling the latency thru the native system and not liking it. NOW, if I had never played him directly analog thru the console to his headphones, would he have been OK with monitoring at 32 samples. No doubt. But once feeling the difference, there was no going back - EVEN THOUGH he knew it was going to cost him thousands of dollars. End of story - no math involved. Only "FEEL". There is a huge difference between "acoustic" latency (or more properly ambient delay) that gives your brian clues about size, depth, dimension, etc. and a latent delay in a pair of cans coming back late when you strike a drum. Big, big difference. It's not only about math my friend.

PS - if there is such a thing as "perfect time" or "perfect groove" (ala perfect pitch) this guitarist has it. His parts are know to be seriously in the pocket and you've heard him thousands of times on the radio. Trust me, there's more going on here than just math.

[matt thomas]

Originally Posted by narco

at those latencies the daw will actually be MORE in time than listening to the drums acoustically from the drummers perspective.

...

just out of interest

narco

[drBill]

Originally Posted by narco

narco

Like I said, nothing to do with math. All about feel. You can't explain art with science. Sorry. Latency matters.

[Fredrik]

I´d say that the brain is capable of detecting delays under one ms atleast when there is a reference. Try this, put on a pair of headphones. Put two identical sounds on two mono channels, pan them hard L,R. Insert a sample delay on one of the channels and start delaying while your listening.

[matt thomas]

Originally Posted by drBill

Like I said, nothing to do with math. All about feel. You can't explain art with science. Sorry. Latency matters.

I repeated it because you missed my point.

I didn't say latency didn't matter. I didn't say its all about maths. I didn't say feel is not important. I didn't say that moving something by 40 samples cant be distinguished

all I said was that the delay through the daw is shorter than from the drums to the drummer.

This is a fact. I posted it out of interest.

apologies to everyone else for getting sidetracked here.

narco

[matt thomas]

Originally Posted by synthoid

I'm a keyboard player, and I've learned over time to play soft keyboard sounds that have huge latencies, so long as the jitter is very low. I've been really surprised at what apparently happened inside my brain, haha, as this happened. I've gotten so that I don't realize that there is a latency -- my hands compensate for it automatically, and I can align my playing with other tracks with good accuracy -- the same as when there is no latency.

-synthoid

this is a good point. however in cases where your brain adjusts to the latency (much like your brain adapts to having your guitar amp and drummer at some distance from your ears) you should be careful. Because if you record such a latency compensated performance then your DAW might also try to compenstate for the latency, so in effect it will be compensating for something that is already compensated for, hence putting it out of time by the latency amount in the opposite direction.

narco

[drBill]

Originally Posted by narco

I repeated it because you missed my point.

I didn't say latency didn't matter. I didn't say its all about maths. I didn't say feel is not important. I didn't say that moving something by 40 samples cant be distinguished

all I said was that the delay through the daw is shorter than from the drums to the drummer.

This is a fact. I posted it out of interest.

apologies to everyone else for getting sidetracked here.

narco

I guess I don't get it- whatever it is your tryng to say. A 32 sample delay via conversion in headphones is much more "noticable" and disruptive than the time it takes for a snare hit to reach the drummers ears. Even though that is longer. If you can't hear it, that's fine. Some can and are disturbed by it. Other's can't tell the latency with 128 samples. Much like some people can't hear when they are singing out of tune. It's a sensitivity that some have and obviously others don't.

For the OP - for many, 32 samples is PRECEIVEABLE latency and it bothers them. For others, it's fine.

[Joe Porto]

I think there is a difference between noticing latency and allowing latency to hinder a performance. Keyboard players have been dealing with latency since the introduction of MIDI as well as digital synthesis and sampling.

I've worked with drummers who complain about 3ms of latency. I've also worked with drummers who complain that one of their 8 toms isn't loud enough in the cue mix. These are the guys who generally don't get called back.

The good drummers just shut up and play, and at worst, may pull one side of the cans off their ear.

[nativeaudio]

He asked me to move a bass part 40 samples. I'm thinking, WTF, there's no way you can hear 40 samples, so I did what any self respecting engineer would do, I said I moved it before I did. He could tell and busted me on the spot. I did some blind A/B tests for him. 9 times out of 10 he could tell.

Sure - and, unlike what some people think, using the 32 buffer in a DAW doesn't mean that you'll get a 32 sample latency. In a Symphony system, the roundtrip latency (using the 32 buffer) @ 44.1 is 148 samples according to the TapeOp review. 148 samples is still low, but my point is that while moving a recorded region 40 samples in a DAW represents only 0.9 ms (40:44.1) difference, using the 32 buffer doesn't mean 0.7 ms (32:44.1) latency. So, the question isn't really if they can hear 0.7 or 0.9 ms latency, but if they have a problem with latencies in the 2-3 ms range, which is what most native and non-native DAWs (including PT HD) can offer.

[SuperTprus]

I play guitar, and I made an experiment to find out how much latency I can percieve some month ago.

For up to 8 milliseconds I don't percieve anything unusual. If I raise the latency above that it starts to feel strange while playing. I think the point is, that I not only hear the sounds, but I feel the vibrating stings as well. It's not a big difference but it's there.

Somewhat like playing with invisible gloves on.. It distracts for a moment but it's easy to compensate.

I know - it is the same latency as standing 3 meters away from the speakers. I think such natural latency is something we've learned to expect but it is a different story if you have the same latency over headphones.

Nils