Digital Igloo (Eric Klein, YGG)

[Eagle]

I agree that clean and edge of breakup tones/feel are harder to nail.

A way to go on these still. I got a pod when it came out but returned it because I thought it must be faulty. It wasn’t it was just pretty bad compared to the outlandish marketing claims.

 

[Digital Igloo]

A way to go on these still. I got a pod when it came out but returned it because I thought it must be faulty. It wasn’t it was just pretty bad compared to the outlandish marketing claims.

I said they're harder to nail, not that we haven't nailed them to both our and our golden ear friends' satisfaction. They get put through the A/B/X wringer as well, and they also landed around 49.5% on our earliest online listening tests, which is close to perfect. A few of our mid-gain amps have even passed null tests (if you compensate for the 1.8ms of latency, of course).

If a Brayden doesn't like a particular HX amp, they simply don't like the sound or feel of the original amp we modeled. Or they're expecting post-processing that may appear on famous records but are clearly not accurate to the original. Any Stan will be able to qualify their statements with nuance.

 

[Eagle]

I said they're harder to nail, not that we haven't nailed them to both our and our golden ear friends' satisfaction. They get put through the A/B/X wringer as well, and they also landed around 49.5% on our earliest online listening tests, which is close to perfect. A few of our mid-gain amps have even passed null tests (if you compensate for the 1.8ms of latency, of course).

If a Brayden doesn't like a particular HX amp, they simply don't like the sound or feel of the original amp we modeled. Or they're expecting post-processing that may appear on famous records but are clearly not accurate to the original. Any Stan will be able to qualify their statements with nuance.

So your idea of a test is adding 1.8ms of digital delay to the analog amp??? Also the analog amp needs its own cab not a monitor. It’s not about the sound as much as the feel.

 

[Lysander]

So your idea of a test is adding 1.8ms of digital delay to the analog amp???

Dude.

 

[Eagle]

Dude.

The main problem is the ADDA.

 

[Digital Igloo]

So your idea of a test is adding 1.8ms of digital delay to the analog amp??? Also the analog amp needs its own cab not a monitor. It’s not about the sound as much as the feel.

We do alllllll sorts of tests. Double-blind A/B/X tests have the original matched cab always active for both the amp (so it's loaded properly) and model. Null tests measure output signals from the preamp outputs, power amp outputs, and mics in front of said cab. Studio monitors are sometimes used, but the real test is in front of the cab in the live room. Success for us is if the model sounds and feels somewhere between two different amps of the same make and model (and ideally, same year); if it's outside of those two, we've failed and we keep tweaking. Sometimes we'll scrap it and start over.

You know this, but for everyone else, a null test determines how close two signals are by flipping the polarity of one signal and adding them together. The resultant signal (measured in level down from 0dB, which includes any A/D and D/A conversion) determines how close you got, and if you're —XXdB or more, Eric Johnson's pet bat can't claim one of the two sounds better—they're effectively, measurably, mathematically identical—again, if you compensate for the 1.8ms of latency, which is the equivalent of standing 18 inches farther from the cab.

Null tests also account for feel, because there's no way two identical waveforms feel different, unless you're talking a playback level mismatch, which is user error, not the fault of a modeler or modeling.

I'm not saying that all Helix amp models are accurate to a fault, nor that all of them pass the null test, and certainly not that they all sound objectively better than anyone else. Only that if someone claims "Oh, Helix models aren't accurate, or modeler X is more accurate than Helix," they're more than likely expecting (or wanting) something other than accuracy, like hyped/scooped models that sound more like a record. Or highly compressed models that feel super saggy and responsive under the fingers. Or the tightness you can get by adding extra fake EQ stages pre and post-preamp. And that's totally fine, but again, it's the opposite of accurate.

 

[KFF]

Cool convo. Great insights. Thanks a lot.

 

[Eagle]

We do alllllll sorts of tests. Double-blind A/B/X tests have the original matched cab always active for both the amp (so it's loaded properly) and model. Null tests measure output signals from the preamp outputs, power amp outputs, and mics in front of said cab. Studio monitors are sometimes used, but the real test is in front of the cab in the live room. Success for us is if the model sounds and feels somewhere between two different amps of the same make and model (and ideally, same year); if it's outside of those two, we've failed and we keep tweaking. Sometimes we'll scrap it and start over.

You know this, but for everyone else, a null test determines how close two signals are by flipping the polarity of one signal and adding them together. The resultant signal (measured in level down from 0dB, which includes any A/D and D/A conversion) determines how close you got, and if you're —XXdB or more, Eric Johnson's pet bat can't claim one of the two sounds better—they're effectively, measurably, mathematically identical—again, if you compensate for the 1.8ms of latency, which is the equivalent of standing 18 inches farther from the cab.

Null tests also account for feel, because there's no way two identical waveforms feel different, unless you're talking a playback level mismatch, which is user error, not the fault of a modeler or modeling.

I'm not saying that all Helix amp models are accurate to a fault, nor that all of them pass the null test, and certainly not that they all sound objectively better than anyone else. Only that if someone claims "Oh, Helix models aren't accurate, or modeler X is more accurate than Helix," they're more than likely expecting (or wanting) something other than accuracy, like hyped/scooped models that sound more like a record. Or highly compressed models that feel overly saggy under the fingers. Or the tightness you can get by adding extra fake EQ stages pre and post-preamp. And that's totally fine, but again, it's the opposite of accurate. And it's the kind of stuff Braydens say, not Stans.

Nothing passes the null test.

 

[Digital Igloo]

Nothing passes the null test.

Not sure what you mean by "nothing," but entire industries are based on passing the null test—live sound, manufacturing safety, automotive audio, noise-cancelling headphones, urban infrastructure... "Passes" means different things to different people, but no human on the planet is going to reliably pass an A/B/X test between two sources that null down to a certain level.

Hell, peruse any of the NAM capture threads and they're constantly talking about passing (or failing) null tests.

 

[Eagle]

Also if your graph tests X&Y at a certain regulatory and this is then rendered identically in the model at this same resolution the results will be identical from the test. This is not the whole story just a test designed to ensure the samples are the same as the analog at these specific points. The resolution then becomes the issue. You yourself say that we are a long way short of sufficient resolution to make this indiscernible from the source.

 

[Eagle]

Not sure what you mean by "nothing," but entire industries are based on passing the null test—live sound, manufacturing safety, automotive audio, noise-cancelling headphones, urban infrastructure... "Passes" means different things to different people, but no human on the planet is going to reliably pass an A/B/X test between two sources that null down to a certain level.

Hell, peruse any of the NAM capture threads and they're constantly talking about passing or failing null tests.

Fine, so we are at modelling is mostly ok with a bit of latency and in a mix which is where I am with it.

 

[Frodebro]

Fine, so we are at modelling is mostly ok with a bit of latency and in a mix which is where I am with it.

I think it depends entirely on what your individual goals and expectations are. For me, modeling as a complete package (meaning the entire signal chain) gets me all the way to what I am after. And I can do that with my Axe FX 3 just as well as my Stomp XL (though obviously each requiring a different approach).

 

[ragingplatypi]

if you compensate for the 1.8ms of latency, which is the equivalent of standing 18 inches farther from the cab.

Can we put this on a sticky somewhere?

 

[Digital Igloo]

Also if your graph tests X&Y at a certain regulatory and this is then rendered identically in the model at this same resolution the results will be identical from the test. This is not the whole story just a test designed to ensure the samples are the same as the analog at these specific points. The resolution then becomes the issue. You yourself say that we are a long way short of sufficient resolution to make this indiscernible from the source.

It depends on the source. As mentioned, some sources (clean, mid-gain, and high-gain) are easy to nail, others (clean, mid-gain, and high-gain) are more difficult and require more DSP. But it also depends on the listener. Put a Brayden in our studio (99.5% of users) and they'll fail A/B/X every time. Put a Stan in our studio (0.5% of users), they'll fail A/B/X most of the time.

The problem is that 100% of users on gear forums think they're a Stan.

Fine, so we are at modelling is mostly ok with a bit of latency and in a mix which is where I am with it.

No. In certain cases with certain models, modeling (as well as ML-based capture tech like NAM) can be effectively identical to the original tube amp. Expensive lab measurement and null tests do not lie; however, our stupid ears lie to us all the time.

When I say a particularly squirrely amp might (???) require upwards of 300% more DSP in the future, that doesn't mean Helix Core 3.80 gets us 33% of the way there. It might get us 96-97% of the way there right now, and that extra DSP may be required to get us to 99.9%. But it may not matter because almost everyone who complains about modeling isn't complaining about the missing 3-4%—they're complaining that modeling can't make their cheap PA speaker behave like a 4x12, or that a hyper-accurate null-test-passing model doesn't sound like what they remember it sounding like on their favorite record. They just say modeling is inferior in nebulous, hand-wavey ways, and you can hear aaalllll about it on their YouTube channel. Don't forget to like and subscribe!

There's no appreciable deficiency in modeling technology, there's an appreciable deficiency in understanding and context. But people reaaaaalllly don't like hearing that because it's easier to blame the tools than it is to learn how to use those tools in a manner that gets them the results they want.

 

[Eagle]

It depends on the source. As mentioned, some sources (clean, mid-gain, and high-gain) are easy to nail, others (clean, mid-gain, and high-gain) are more difficult and require more DSP. But it also depends on the listener. Put a Brayden in our studio (99.5% of users) and they'll fail A/B/X every time. Put a Stan in our studio (0.5% of users), they'll fail A/B/X most of the time.

The problem is that 100% of users on gear forums think they're a Stan.

No. In certain cases with certain models, modeling (as well as ML-based capture tech like NAM) can be effectively identical to the original tube amp. Expensive lab measurement and null tests do not lie; however, our stupid ears lie to us all the time.

When I say a particularly squirrely amp might (???) require upwards of 300% more DSP in the future, that doesn't mean Helix Core 3.80 gets us 33% of the way there. It might get us 96-97% of the way there right now, and that extra DSP may be required to get us to 99.9%. But it may not matter because almost everyone who complains about modeling isn't complaining about the missing 3-4%—they're complaining that modeling can't make their cheap PA speaker behave like a 4x12, or that a hyper-accurate null-test-passing model doesn't sound like what they remember it sounding like on their favorite record. They just say modeling is inferior in nebulous, hand-wavey ways, and you can hear aaalllll about it on their YouTube channel. Don't forget to like and subscribe!

There's no appreciable deficiency in modeling technology, there's an appreciable deficiency in understanding and context. But people reaaaaalllly don't like hearing that because it's easier to blame the tools than it is to learn how to use those tools in a manner that gets them the results they want.

Thanks for taking the time. I use digital all the time and am mostly ok with it but my tube gear feels better and that is my main issue.

 

[Digital Igloo]

Thanks for taking the time. I use digital all the time and am mostly ok with it but my tube gear feels better and that is my main issue.

So this might be a good time to discuss actual problems with modeling. First of all, it requires a lot more understanding and nuance!

Reaching over and turning up the Drive knob on a non-master-volume tube amp is a wholly different experience from needing to hook up multiple pieces of gear and then adjust multiple knobs (where applicable): Input Level, Model Gain, Model Channel Volume, Output block Level, Modeler Main Volume, playback system Input Trim, playback system Volume, etc. There's absolutely additional know-how and tweaking required to get modelers to behave like amps, and the laws of physics dictate this, not any lack of DSP resources or clever programming.

So why not get rid of all that extra crap?

Well, Fender does a good job with their standalone Tone Master amps, but they work in large part because they're not trying to be all things to all people—they're just one particular Fender amp, albeit with a modeled preamp and solid state power amp. With multieffects, you're always on the lookout for things that'll undermine the goal of nailing the sound and feel of a particular amp. (Provided that is the goal; it may not be.)

Modeling-based setups also homogenize the sound of amps, because whatever playback system is used—even a tube power amp and cab—the various amp models are often level-matched to avoid drastic volume changes. And they're all coming out of the same speaker. The default Master volume in each model is also often set fairly high because that's the real amp's sweet spot. But hearing that from a playback system at a much lower volume sounds and feels totally off.

So if someone says "Modeling technology isn't accurate," science can prove they're wrong. But if someone says "It's difficult to get modelers to sound and feel accurate" or "Modeling can be a pain in the ass," I'll shrug and say "Yeah, they can be." But for many people, the advantages are worth the effort.

 

[Eagle]

So this might be a good time to discuss actual problems with modeling. First of all, it requires a lot more understanding and nuance!

Reaching over and turning up the Drive knob on a non-master-volume tube amp is a wholly different experience from needing to hook up multiple pieces of gear and then adjust multiple knobs (where applicable): Input Level, Model Gain, Model Channel Volume, Output block Level, Modeler Main Volume, playback system Input Trim, playback system Volume, etc. There's absolutely additional know-how and tweaking required to get modelers to behave like amps, and the laws of physics dictate this, not any lack of DSP resources or clever programming.

So why not get rid of all that extra crap?

Well, Fender does a good job with their standalone Tone Master amps, but they work in large part because they're not trying to be all things to all people—they're just one particular Fender amp, albeit with a modeled preamp and solid state power amp. With multieffects, you're always on the lookout for things that'll undermine the goal of nailing the sound and feel of a particular amp. (Provided that is the goal; it may not be.)

Modeling-based setups also homogenize the sound of amps, because whatever playback system is used—even a tube power amp and cab—the various amp models are often level-matched to avoid drastic volume changes. And they're all coming out of the same speaker. The default Master volume in each model is also often set fairly high because that's the real amp's sweet spot. But hearing that from a playback system at a much lower volume sounds and feels totally off.

So if someone says "Modeling technology isn't accurate," science can prove they're wrong. But if someone says "It's difficult to get modelers to sound and feel accurate" or "Modeling can be a pain in the ass," I'll shrug and say "Yeah, they can be." But for many people, the advantages are worth the effort.

In my tests I have done my best to counter as much of this as possible and indeed had the most successful results in as far as as behaving like the source amp with a Kemper profile made by me then without moving it play the profile and the amp ABX through the same cab still in situ.

 

[Digital Igloo]

In my tests I have done my best to counter as much of this as possible and indeed had the most successful results in as far as as behaving like the source amp with a Kemper profile made by me then without moving it play the profile and the amp ABX through the same cab still in situ.

Yep, and as soon as you want to change amps (or change playback volume because your ears start to ring), you often gotta do more work. Chasing sound/feel accuracy is totally doable in digital, but it's not necessarily fun. That's why I've been gravitating toward Line 6 Original amp models: lower DSP (because they don't need to describe weird, perhaps unpleasing behavior), more forgiving, and no has-this-setup-been-tweaked-to-sound-as-accurate-as-possible voice in your head.

 

[KFF]

So if someone says "Modeling technology isn't accurate," science can prove they're wrong. But if someone says "It's difficult to get modelers to sound and feel accurate" or "Modeling can be a pain in the ass," I'll shrug and say "Yeah, they can be." But for many people, the advantages are worth the effort.

Also probably why the Cooper Carter's of the world are employed by large acts to get things straight.

Sorry for cross-pollination to FracalLand... its all I know these days. But 20 something years ago, the original Amp Farm TDM changed my life. Then Echo Farm came along and changed it even more! Based on your carry-on ( ) here today, I feel I simply must visit L6. Where do I dip the toe? Stomp?

 

[ragingplatypi]

Where do I dip the toe? Stomp?

yup