Latency of amp modelers

so run some tube amps? 🥹
No. Run headphones or in-ears. You can substract some ms of latency from the sound not going some feet through the air anymore, but you have to add some ms latency for the processing and then your latency is on spot again. You stand 3 feet away from your rack and the sound comes after 3ms, now that's weird...
 
...

This is the problem with the internet and videos like in the OP. Now latency has somehow become the most important feature of an amp modeler even though it's been long established that most people cannot detect anything below 10ms and even the most attuned performer can't detect below 5ms. And people are determined to "prove" that sub-ms latency matters and go to great lengths to find evidence and use papers as evidence of this even though the papers have nothing to do with I/O latency.

We now have two papers cited in an attempt to prove that latency below 5ms is detectable and neither of these papers is at all relevant.
I think, latency comparisons have become more prominent after many of today's modelers have become really good in terms of tone and while tone is an extremely subjective topic, latency is objectively measurable, whether it's trivial to measure or not.

Tests where people are trying to distinguish between two acoustic events (like snare and hihat) offset by few milliseconds IMHO is not comparable to the sensation of latency when picking a string and hearing the processed acoustic response. I have been messing around with the Quad Cortex for a few days and being used to play through headphones I have a feeling of disconnection between my playing and the headphone sound with the QC. I had a similar feeling with the Kemper. It's somehow weirdly obvious on these units while not so obvious on some other modelers. Even the NUX MG-30 feels more "connected" to my play. Interestingly it's one of the units with relatively low latency in Leo's test. Also, kudos to BOSS for achieving a good feel even in their Waza Air system despite the wireless link.

I guess, latency is more perceptible in certain scenarios and when it is, it's really annoying. Remember, it's not only about just hearing a difference between two sounds. It's about expecting to hear a response to something, you feel with your (picking) hand. Maybe it's a lesser problem when monitoring in a room, but with headphones it can really suck.
 
No. Run headphones or in-ears. You can substract some ms of latency from the sound not going some feet through the air anymore, but you have to add some ms latency for the processing and then your latency is on spot again. You stand 3 feet away from your rack and the sound comes after 3ms, now that's weird...
You missed my joke. Tube amps can double as space heaters.
 
No. Run headphones or in-ears. You can substract some ms of latency from the sound not going some feet through the air anymore, but you have to add some ms latency for the processing and then your latency is on spot again. You stand 3 feet away from your rack and the sound comes after 3ms, now that's weird...
...provided, you have a similar impression of an "in the room sound" through your headphones or in-ears, which most likely won't be the case.
 
Maybe it's a lesser problem when monitoring in a room, but with headphones it can really suck.
The issue is bigger with monitors in a room, since you have some natural latency from the distance of the source to your ears, then add some latency from a device. With headphones you have no natural latency....
 
so run some tube amps? 🥹
I would say, run a tube amp if you have one you really like...... I still run my Princeton Reverb a lot more than the preset on my AF3 - just because it feels great, sounds great and it is much simpler than a device that can do a hundertthousend things I don't need in such a moment......

PS: This is about tools, not about religions. Tube amps are still around for a reason even with high end modelers which get better and better......
 
....
With headphones you have no natural latency....
Exactly, that's why I think, latency is a bigger issue for our brain when playing through headphones, even if it's absolutely lower than in the room. Unless you have headphones that are virtually absent by giving you a perfect impression of a speaker in a room at a certain distance.
Interestingly, panning two amps or different speakers hard left and right in the modeler feels even more "latent" (or disconnected) with headphones although it shouldn't change the absolute latency of the signal.
 
Sorry to beat a dead horse. But this information is more or less impossible to find (meaning actual reported latency measurements from guitar > through Fractal > through a simple amp/cab). No additional wireless units, or external pedals, or FRFR software, not even the latency when used as a USB interface into a DAW - just a simple A/D/A conversion from analog inputs to analog outputs.

My question is this: Is there a measurable difference in throughput latency between the "AxeFX III MkII Turbo" and "FM9 or FM9Turbo"? If so, how many milliseconds? Let's assume we are comparing the same preset, like 65 BASSGUY or something with stock settings.

- I'm not really concerned with CPU power. FM9 has enough power to do what I want. Although, I am currently on the FM9 Turbo waitlist, because there are a few situations where the extra power could help. Even so, it seems that the A/D/A are largely responsible for latency characteristics anyway not necessarly CPU.

- If AFX3mk2Turbo has faster latency, there's a chance it might actually swing my purchasing decision. Will it be worth another $800 to me? Probably not, but it's a little frustrating that I can't seem to find the numbers, lol.

///

Also to be clear, I agree with everyone in this thread. Some folks say, "I can't feel latency, so therefore you can't either". Other folks say, "I can feel 1ms, so until we get 0ms, we should stick with our old vintage pedals". Then the techies chime in with reason and logic, saying things like "no one perceives latency below 5ms anyway". It's all good, and you're all correct to a degree. But where are the numbers? I just want a couple numbers, lololol.

With some measurements, I can decide whether it's worth an extra $800 to me. (It probably isn't, but curiosity is killing the cat, so to speak).

Thanks all! You guys rule. I don't mean to come off as a "rant". It's just been eating at me for a while. We're all so quick to point out exactly how fast sound travels through air, why is the speed through the device such a secret? Personally, I think Fractal should post some numbers and be proud of it - man, to deliver these sounds, as fast as they do? Dang, that should be a feature not a secret.

Who knows, maybe I just haven't googled the right combination of words to land on the master list of modeller latency specs. Haha.
 
Sorry to beat a dead horse. But this information is more or less impossible to find (meaning actual reported latency measurements from guitar > through Fractal > through a simple amp/cab). No additional wireless units, or external pedals, or FRFR software, not even the latency when used as a USB interface into a DAW - just a simple A/D/A conversion from analog inputs to analog outputs.

My question is this: Is there a measurable difference in throughput latency between the "AxeFX III MkII Turbo" and "FM9 or FM9Turbo"? If so, how many milliseconds? Let's assume we are comparing the same preset, like 65 BASSGUY or something with stock settings.

- I'm not really concerned with CPU power. FM9 has enough power to do what I want. Although, I am currently on the FM9 Turbo waitlist, because there are a few situations where the extra power could help. Even so, it seems that the A/D/A are largely responsible for latency characteristics anyway not necessarly CPU.

- If AFX3mk2Turbo has faster latency, there's a chance it might actually swing my purchasing decision. Will it be worth another $800 to me? Probably not, but it's a little frustrating that I can't seem to find the numbers, lol.

///

Also to be clear, I agree with everyone in this thread. Some folks say, "I can't feel latency, so therefore you can't either". Other folks say, "I can feel 1ms, so until we get 0ms, we should stick with our old vintage pedals". Then the techies chime in with reason and logic, saying things like "no one perceives latency below 5ms anyway". It's all good, and you're all correct to a degree. But where are the numbers? I just want a couple numbers, lololol.

With some measurements, I can decide whether it's worth an extra $800 to me. (It probably isn't, but curiosity is killing the cat, so to speak).

Thanks all! You guys rule. I don't mean to come off as a "rant". It's just been eating at me for a while. We're all so quick to point out exactly how fast sound travels through air, why is the speed through the device such a secret? Personally, I think Fractal should post some numbers and be proud of it - man, to deliver these sounds, as fast as they do? Dang, that should be a feature not a secret.

Who knows, maybe I just haven't googled the right combination of words to land on the master list of modeller latency specs. Haha.
Did you see Cliff's post on page 2?
 
  • Like
Reactions: Rex
He has six posts on that page. All worth reading.
This whole thread is jam packed with great info. But will fm9 have the same number as fm3? One person said that fm9 would fall between fm3 and afx3, but they were quickly dismissed as incorrect.
 
The FM9 has the same DSP as the FM3, but there's two chips instead of just one (4 cores vs 2 cores of the FM3). The turbo version uses chips that run at a slightly higher clock rate (500MHz vs 450MHz). I would expect its latency specs to be pretty similar to the FM3, maybe a tiny bit better due to the faster clock speed. There's only about 1 ms latency difference between the Axe III and the FM3, so the whole latency thing is basically a non-issue in that respect. That's the equivalent of moving about 1 foot further from your speakers, which you will almost certainly never notice.
 
As I said before, my interest lies in the question as to whether 5 ms is a definitive latency floor. My reading of the Lester paper says it isn’t and I take it that 5 ms has been long used as a useful engineering rule of thumb. The following chart from the paper characterizes by instrument the range of latencies the subjects found acceptable. To me this was non-intuitive. I wouldn’t have expected different instruments to demonstrate such a varied range of manageable latencies.
View attachment 104645

It makes sense to me: a singer is directly producing sound with their voice (close to their ears and felt/heard directly) can probably detect latency better. A horn player might be similar as the instrument is very close and loud.

For a digital keyboardist playing there is no direct audible feedback (and possibly digital latency anyway that they are used to). Even with a piano, there is a distance to the strings and resonance takes time to develop in the sound board.

Just a thought.
 
Last edited:
You can put an effects loop in a preset and shortcut that external loop with a patchcable and switch the loop on and off and that way you get more or less latency in that preset and you can try if it matters or not. Be sure to have exactly the same volume in both cases, because even tiny differences make a different impression.
 
Back
Top Bottom