Latency - What do you hear?

Roland

The thing which has always irritated me about digital audio processing is latency. Whenever I add Chorus or Detune I hear his metallic sound. Now I know why, even at 3ms, it irritates me.

https://youtu.be/3zZRy-UArXM?si=zn8FmlkGYR4u__ND

Am I in a world of my own, or can you hear it too?

andy_k

Hmm,
I think latency is more of a problem when you are trying to perform with it, the sonic effect is just that, an effect, and is more of a phase problem.
Not limited, or strictly related to digital, ie tape flanging, / chorussing, or just bad maintenance.
FWIW, I think I can hear 1ms in your example, compared to 0ms , but I doubt it would affect my performance, and if it did, I could always move it back into phase, that is the beauty of digital, you can see the waveform at a tick level.

Keefy

The speed of sound in dry air at 20C is 343m/s, so it takes sound 2.9ms to travel 1m. If you move from a position 2m in front of your amp, to say 5m away (big stage), that’s an added latency of 8.7ms. Other than the volume, do you notice a difference? I’m guessing not.

But if your guitar sound is coming to you direct and via a digital processor with even a small amount of latency, there will be phase cancellation at certain frequencies and you’ll likely hear a ‘comb filtering’ effect which may well be unwanted and unpleasant.

Danny1969

Keefy said:

The speed of sound in dry air at 20C is 343m/s, so it takes sound 2.9ms to travel 1m. If you move from a position 2m in front of your amp, to say 5m away (big stage), that’s an added latency of 8.7ms. Other than the volume, do you notice a difference? I’m guessing not.

But if your guitar sound is coming to you direct and via a digital processor with even a small amount of latency, there will be phase cancellation at certain frequencies and you’ll likely hear a ‘comb filtering’ effect which may well be unwanted and unpleasant.

I  certainly do notice distances like 5m on stage, it's quite a lag. I always have a lot of hi hat in the wedge or on my ears on medium stages let alone big stages because that's the only way you are going to be locked tight with  the drums.

One thing you notice about old gigs with huge crowds is the bands often stayed together quite close on stage even when the stages were big. I guess the monitoring wasn't great and that solved that problem. 
 

snowblind

Depends to some extent on the gear you are using. I've noticed with some of the cheaper digital amps there is a noticeable lag between hitting the strings and the noise coming out of the speaker. It's like watching a film with the audio out of synch. It gets very annoying once you notice it and it's hard to ignore it once you become aware. The easy solution is to avoid cheap digital amps of course.

ICBM

Roland said:

The thing which has always irritated me about digital audio processing is latency. Whenever I add Chorus or Detune I hear his metallic sound. Now I know why, even at 3ms, it irritates me.

https://youtu.be/3zZRy-UArXM?si=zn8FmlkGYR4u__ND

Am I in a world of my own, or can you hear it too?

That’s not actually latency you’re hearing, it’s the combination of a signal with latency, with one without. Latency on its own is much harder to tell when it’s in the few-ms range - you can’t actually hear it (because there’s no difference in the sound), it’s just that as it gets larger you become aware of the lag.

Combining a signal with latency with one without - eg in a parallel FX loop where the effect unit in the loop is digital and has latency (unless it’s purely a delay or reverb set to wet-only) always sounds bad at any amount of latency, even very short, because it causes a comb-filtering effect, like a static flanger.

Roland

ICBM said:

That’s not actually latency you’re hearing, it’s the combination of a signal with latency, with one without.

This arose when I first got hearing aids. I was hearing both the original sound and the frequencies amplified by the aids. When I put a plate on the table it sounded cracked. I had the comb filtering conversation with the audiologist, who surprisingly had never encountered it before. We turned off most of the clever processing features in the hearing aids to reduce the problem. After two weeks my brain had adjusted, and I didn’t notice the problem any more. However it’s still there with FX.

ICBM

I first found it a problem with a Mesa amp with their notoriously poor parallel FX loop - with an all-digital FX unit in the loop (a Line 6 DL-4, one that digitises the dry signal as well) it sounded terrible even with the loop mix at 100% - because the mix is just a crude single pot, and ‘100%’ is actually only about 90% - the small amount of direct signal remaining was enough to produce that metallic comb-filtering sound. I fixed it simply by converting the loop to pure series, at which point the signal was now *all* delayed (I think it was also about 3mS from memory - I did measure it) and sounded fine.

Musicwolf

I play using a digital modeller, often via a digital mixer, so there will always be some latency.  That's my baseline.

I own a Nux B-7PSM digital in-ear transmitter which quotes latency as <5.5ms (I'm taking this at face value, I've not tested it).  I also own a couple of wireless guitar transmitters.  A Line 6 G30, latency <2.9ms, and a Lekato WS-60 stereo transmitter <12ms.  I tried the Lekato because I have a couple of piezo equipped guitars with twin outputs.

Using the Nux (<5.5ms) or the G30 (<2.9ms) only - OK.

Using the Lekato (<12ms) only - Horrible.

I can't recall whether I've tried the Nux and G30 in combination but I have tried offsetting a guitar track in my DAW and I can really hear it at 10ms.  If I get bored I may run a little experiment later to find my threshold value.  The problem is that it should really be a double blind test and I don't believe for a moment that anyone else in the house is remotely interested in helping me.

Sporky

The video is odd for me. 0.5 to 2.0 sound the same as each other, then I can hear both. 

stratman3142

I can hear a phasing type effect for the low settings. Then there are two distinct sounds as the delay increases.

I used to have a rack based live rig with a t.c. electronics G Major processor in the effects loop of a Marshall JMP1 pre-amp.  I only used the time based effects in my the G major processor (e.g. delay reverb, chorus etc) and ran it 100% wet, otherwise I could hear a phase cancellation type effect. I believe it was referred to as a pipeline delay effect.

For me, that's different to the perception of latency when recording with an audio interface. As latency increases, I start to 'feel' a lack of instantaneous response (i.e. a difference in plucking a string and the sound it produces).

Typically I record using headphones with an input/output latency as 1.4/4.0 (i.e. a total of about 5.4ms) with my Zoom UAC 2 interface. I can't feel that. Sometimes I've had to increase that to ease the load on my computer when running loads of plug-ins. I start to feel a lack of instantaneous response (a lag) somewhere in the region of 8ms, although I can cope with a bit higher.

The above is under the microscope of recording.  Live I can cope with a bit higher, but I notice a lag if I get too far from my amp or a monitor.

TheBigDipper

Danny1969 said:

One thing you notice about old gigs with huge crowds is the bands often stayed together quite close on stage even when the stages were big. I guess the monitoring wasn't great and that solved that problem. 
 

I thought it was just because the band liked being close to each other to communicate during the performance. Spreading out across the stage kills that.  I loved the way Led Zeppelin stayed close when they did the O2 gig. You don’t need to spread out on the age of cameras and screens. 

fretmeister

Every time I start to wonder about it I realise that once the drums start and the snare echoes round the room I can't hear it anymore anyway.

Roland

Sporky said:

The video is odd for me. 0.5 to 2.0 sound the same as each other, then I can hear both. 

To my ears the very short delays (up to 3ms) aren’t identifiable as two notes, but as differences in the timbre of the note.

fretmeister said:

… once the drums start and the snare echoes round the room I can't hear it anymore anyway.

It doesn’t work that way to my ears. 

I’ve come to think that differently people’s brains might process sound differently. That’s whi I asked the question. We assume we are similar, and that might not be the case.

Sporky

Roland said:

Sporky said:

The video is odd for me. 0.5 to 2.0 sound the same as each other, then I can hear both. 

To my ears the very short delays (up to 3ms) aren’t identifiable as two notes, but as differences in the timbre of the note

Me too, those three sound different from no delay, but not much different from each other. 3.1ms onwards I can hear two. 

Nerine

Roland said:

The thing which has always irritated me about digital audio processing is latency. Whenever I add Chorus or Detune I hear his metallic sound. Now I know why, even at 3ms, it irritates me.

https://youtu.be/3zZRy-UArXM?si=zn8FmlkGYR4u__ND

Am I in a world of my own, or can you hear it too?

How is adding chorus or detune letting you “hear” latency?

Latency is just a uniform time delay of the whole signal — you don’t hear that as a tonal change unless you’re monitoring against the dry input.

Chorus and detune do use very short time delays, but those are built into the effect itself. What you’re hearing is the delayed version blended with the dry, which creates phase interactions (comb filtering). That’s what gives it that metallic character.

If you import two identical tracks into a DAW and nudge one a few samples, you’ll hear the same thing - not latency per se, but phase cancellations. I can notice this at a few samples, let alone a few milliseconds.

So the video is a better demonstration of comb filtering/phase alignment than of latency, in my opinion. 

Phase shift is what you’re hearing up until there is an obvious repeat of the transient. 

It’s one of the reasons we make sure microphones are “in phase” (and usually measured) before recording anything - especially drums - because it’s very easy to make a multi-mic’d drum kit (for example) sound like shite if the microphone distances and phase relationships aren’t aligned properly. 

Out of phase sounds usually have a hollow or nasal quality to them. Often it affects low end more because the wave lengths are longer and therefore there is more opportunity for something else to be out of phase or interfere with it. 

It’s also one of the reasons I’ll use a mono overhead a lot of the time, live. Much easier to get sounding coherent than a pair when time is limited. 

Roland

Nerine said:

How is adding chorus or detune letting you “hear” latency?

Mixing the two signals means that I can hear the effect of latency. One of the techniques for fattening a guitar sound is to combine the original signal with detuned signals. Between +7% and +11% are often used.