Real amps vs modelling and plugin amps

[Magic Russ]

GuitarRig has a “High Quality” button if I recall which may be oversampling not sure.

They removed the High Quality button in version 6, making that the default (only) mode. It seems HQ mode was 2x oversampled in GR5, but I don't know if they increased the oversampling in later versions.

[stoopicus]

I just did a survey of all the distortion boxes and high gain amps in Amplitude, and it was fascinating how uneven it is. Many show no signs of aliasing at all, some are pretty bad. I just ran a sine wave from Massive and played a C7 into it and used the pitch wheel to bend it up. No cabs. Even amps in premium packs had a wide range of quality. For instance, the Mesa Mark IIC+ Lead channel? Aliases like a stuck pig. Mark III Lead channel? None. Satch VM? Aliasing galore. Satch Distortion? Clean. I wasn't taking notes, but I should do it again and make a list. I didn't check to see if the Tonex stuff aliased, but I do remember noticing the NAM plugin did.

Yes, would be great if you can list everything! I can send it to IK, if you don't mind.

Also, if you'd like to run a test like that for Tonex, I can send you a license for the Boutique Overdrives expansion.

I had some time today, so I went through all the models.

Just as a note to save yourself difficult qualitative listening, if you use a frequency-domain scope like an EQ plugin, with a sine wave as input even after processing you should be able to easily visibly see the aliasing reflected in as noise coming down from the Nyquist frequency.

A good example is the graph here, where the tall line at 1k is the sine wave fundamental, the taller lines to the right are harmonic distortion, the shorter lines very close to the harmonic distortion lines, coming in from the right and decreasing in magnitude as they get lower, is aliasing; the fuzzy noise likely has some too but might just be noise. Nyquist is at some frequency to the far right, at or beyond 20k.

https://varietyofsound.wordpress.com/20 ... -aliasing/

[zerocrossing]

I had some time today, so I went through all the models.

Thanks! I'll send this list to my IK rep.

I updated it with fuzz models (only two had some moderate aliasing).

I also started going through some of the Tonex models, and here's my take, most are very good and only show a very tiny amount. Still getting a 60 db s/r. Some user high gain amps that didn't have cabinets were worse. Could that be because they were brighter sounds.

I briefly started going through Guitar Rig 7's amps, and straight off the bat the aliasing is worse on some models, like the ACs. Chicago was good, though. Why? No idea. Citrus was weird.. like maybe not aliasing, but something else... Cool Plex, good. Fire Breather, bad.

I got bored of it, so I stopped at Fire Seeker, which was fine on 1, worse on 2 and even worse on 3. So, kind of all over the place, like Amplitube.

I should also note that C6 is a very high note, and when I had cabinets engaged, the aliasing was almost always either filtered out or diminished to a point where you wouldn't notice it. I didn't test any Neural DSP stuff, because I don't own any.

[guitarzan]

Here’s the ToneX aliasing thread on The Gear Page:

https://www.thegearpage.net/board/index ... g.2604401/

I can hear the aliasing pretty clearly on those tests, I don’t know what the signal to noise ratio is, but I know something makes me quit liking ToneX when it starts overdriving more than just the lightest amount (‘driven clean’ tone and feel = excellent, ‘moderate overdrive’ = grating). Oddly, outside of these tests I only notice when I’m playing, not really on most recorded examples, but it’s enough to just make me not go there.

[zerocrossing]

Here’s the ToneX aliasing thread on The Gear Page:

https://www.thegearpage.net/board/index ... g.2604401/

I can hear the aliasing pretty clearly on those tests, I don’t know what the signal to noise ratio is, but I know something makes me quit liking ToneX when it starts overdriving more than just the lightest amount (‘driven clean’ tone and feel = excellent, ‘moderate overdrive’ = grating). Oddly, outside of these tests I only notice when I’m playing, not really on most recorded examples, but it’s enough to just make me not go there.

I couldn't get any audible aliasing on clean, lightly driven or crunch tones on Tonex or Amplitube models. At least not high enough from the noise floor where I could hear it at all. When I reported by findings, they were high gain amps at full gain, presence up. Many high gain models didn't even show any. Some did. I purposely made it a worse case scenario. I'm also playing a note that would be way up the neck. I honestly don't think you're hearing aliasing. Even when I did find it, it was often at about -60 db. The internet is telling me that the average dynamic range of a tube amp is between 50 and 60 db.

https://rubytubes.com/blogs/the-audioph ... amic-range

On top of this, I was listening to naked distortion box and amp emulations. The limited frequency response of a guitar cabinet is going to pad the aliasing down a bit and make it even less noticeable.

[stoopicus]

The internet is telling me that the average dynamic range of a tube amp is between 50 and 60 db.

https://rubytubes.com/blogs/the-audioph ... amic-range

I am legit surprised it is that much

[zerocrossing]

The internet is telling me that the average dynamic range of a tube amp is between 50 and 60 db.

https://rubytubes.com/blogs/the-audioph ... amic-range

I am legit surprised it is that much

Oh, I'm sure that's best case. Real world? Probably less. I had an Ampeg Reverb Rocket that I really liked, but I eventually sold it because it was just so noisy.

[Uncle E]

So this is interesting, the way I use Tonex is by running a capture of my speakers after the capture of the DI'd amp. Since the speaker capture is mostly clean, it wouldn't have any aliasing, and I wonder if it's filtering out the aliasing from the amp capture.

[guitarzan]

… I couldn't get any audible aliasing on clean, lightly driven or crunch tones on Tonex or Amplitube models. At least not high enough from the noise floor where I could hear it at all…

Did you did you get a chance to check out the examples on The Gear Page?
https://www.thegearpage.net/board/index ... t-39840674
There I can clearly hear the aliasing even in the clean Fender Twin ToneX profile.

The sweep used there was created by Jay Mitchell specifically for testing amp sims:

I formulated the sweep Richard G used specifically to contain the same frequency range as guitar, while making it easier to identify aliasing and to distinguish it from other forms of noise. If you can hear aliasing when you use the sweep, it will be audibly present when you play guitar through the same device with the same settings. The only difference will be that, with guitar as input, aliasing will be more difficult to recognize as such. Eliminating aliasing will always improve the sound…

… In the case of this device [ToneX], the first aliased sounds move rapidly upward and downward a la the "Pacman" sound. This means that the aliased tones are high-order harmonics. There is clearly some oversampling, but just as clearly not to high enough order.
https://www.thegearpage.net/board/index ... t-39843382

The aliasing is pretty prominent up top on all examples. Must be some differences in settings or the sweep used in your tests or something?

[zerocrossing]

… I couldn't get any audible aliasing on clean, lightly driven or crunch tones on Tonex or Amplitube models. At least not high enough from the noise floor where I could hear it at all…

Did you did you get a chance to check out the examples on The Gear Page?
https://www.thegearpage.net/board/index ... t-39840674
There I can clearly hear the aliasing even in the clean Fender Twin ToneX profile.

The sweep used there was created by Jay Mitchell specifically for testing amp sims:

I formulated the sweep Richard G used specifically to contain the same frequency range as guitar, while making it easier to identify aliasing and to distinguish it from other forms of noise. If you can hear aliasing when you use the sweep, it will be audibly present when you play guitar through the same device with the same settings. The only difference will be that, with guitar as input, aliasing will be more difficult to recognize as such. Eliminating aliasing will always improve the sound…

… In the case of this device [ToneX], the first aliased sounds move rapidly upward and downward a la the "Pacman" sound. This means that the aliased tones are high-order harmonics. There is clearly some oversampling, but just as clearly not to high enough order.
https://www.thegearpage.net/board/index ... t-39843382

The aliasing is pretty prominent up top on all examples. Must be some differences in settings or the sweep used in your tests or something?

I did listen to those, and what you hear on the Twin is so minimal I wouldn't even consider it. In a real world scenario, you're just not going to hear that. Also, what is that sound they're using? It alone sounds kind of weird and digital, like a sample of a guitar that's being pitched up. Frankly, I don't trust The Gear Page posts at all, because those dudes are basically tube amp evangelists and I'm surprised they don't go around on the weekends knocking on doors, asking if people have heard the good news about tube guitar amps.

I use this stuff all the time. I'm also using synthesizers a lot, and I'm aware when digital synthesizers alias as well. Not just software, but expensive hardware synthesizers, like the Virus TI, as well. If any of this was a serious issue, I simply wouldn't use it. That said, I'm not an ultra high-gain guy, and I almost always begin with a healthy signal level into my Apollo that I then pad down 15 db on the input gain, and go from there. I have to laugh at the Ola Englunds of the world, because whenever I hear them demoing some high gain amp, I think it sounds terrible. I've often wondered why ENGL has ever sold an amp, but as they say, there's no accounting for taste.

In the real world, we have this...



Can anyone listen to that and say, "oh, the horrible aliasing!" Sounds great to me. Better than many demos I've heard of the real deal that weren't made in a pro studio. I had to laugh at some dude's track he posted recently. He was so proud of his all analog guitar/synth studio, and then he posted something that sounded like it was recorded with an old Radio Shack mic he set up outside his studio.

[guitarzan]

That chirpy UFO noise is NOT subtle on those tests, it is clear as day even on the clean ToneX profile. It’s warbling like crazy at the uppermost frequencies.

It really doesn’t bother me when recorded, particularly in a mix, but I now feel certain that I do hear that aliasing while playing it in real life, only I couldn’t identify it as aliasing before those tests.

The upper harmonics are the exact frequencies I key on while playing, it is where the ‘chime’ is at and for me electric guitar is all about ‘chime’. That is why it bugs me while playing but gets a pass from me when recorded. It can be fixed and it should be fixed.

It comes off like bridge noise while I’m playing so I’m constantly trying to adjust my playing (pick attack, etc) to minimize it but nothing touches it.

The main actors in that test in the gear page are all long time amp modeler users, not tube snobs. They are trying to point out a flaw that needs to be addressed. IK telling everybody what they can and can’t hear is why I always avoided them, this product is so close, why not do it right?

[zerocrossing]

So this is interesting, the way I use Tonex is by running a capture of my speakers after the capture of the DI'd amp. Since the speaker capture is mostly clean, it wouldn't have any aliasing, and I wonder if it's filtering out the aliasing from the amp capture.

I wonder what is happening with these captures. When I first started listening to the Tonex captures, I checked a bunch of gain pedals and I wasn't hearing anything prominent, but after a while, I did start coming across some were it was noticeable. It seemed kind of hit-or-miss. Later, I'm going to try my little test on patches that I personally use and think sound really good. I'll report back.

I'm also wondering if my test was possibly skewed because I was using a digital synth plugin. It's a good one, Massive X, but maybe all these tests are showing exaggerated flaws in the source signal. If I have some time later, I'll try the same test with an analog input.

[zerocrossing]

It really doesn’t bother me when recorded, particularly in a mix, but I now feel certain that I do hear that aliasing while playing it in real life, only I couldn’t identify it as aliasing before those tests. The upper harmonics are the exact frequencies I key on while playing, it is where the ‘chime’ is at and for me electric guitar is all about ‘chime’. That is why it bugs me while playing but gets a pass from me when recorded. It can be fixed and it should be fixed.

The weird thing is, I was surprised by a lot of these. They were coming from models that I frequently use, though not at such extreme levels. When I was seeing if I could use Tonex and Amplitube instead of my AxeFX, I did check to see if I was hearing any aliasing. I did this by playing a high note and bending it, and I'm sure I would have noticed any egregious or modest aliasing. I have before. At those high notes, fret buzz is a far bigger issue. There was some Italian plugin that some dude declared was the best amp modeler ever, and I tried it and instantly noticed it aliased a lot. I noticed it on some NAM models too. I'm talking about real world use, not sending synthesizer oscillator tones into it.

Check this out, starting at 7:02. This is all Amplitube, no Tonex, I don't think, (I'll check the file later) both the rhythm and lead sounds. No external gain effects, either. I need to keep things in software because publishers are often making changes and I need to be able to adapt quickly.

https://soundcloud.com/zerocrossing-1/m ... ic_montage

Excuse the cheez factor of the tune, it was for a casual VR game that had a section modeled after an old school boardwalk amusement park, and I was trying to capture an 80s Jersey Shore vibe. (each level of the game had a different theme) My point is, it's fairly high gain, but typical for me, and I don't hear it as being harsh in any way, or having noticeable aliasing.

[guitarzan]

…I'm talking about real world use, not sending synthesizer oscillator tones into it…

The sweep used on the Gear Page test was developed by Jay Mitchell, who is an audio engineer, specifically to represent the frequencies of an electric guitar including the upper harmonics, which is where the problem lies, in those complex upper harmonics—not necessarily limited to high frequency fundamentals. Those frequencies would be in use constantly with a chimey electric tone, even on cowboy chords.

As I already said, it seems minor when recorded, but while I play it irks me like a rattling bridge noise, but nothing can settle it. Grating but not earth shattering. Enough to make me jump through a lot of hoops to avoid it though, when more oversampling and maybe some strategic high cut filtering might very well settle it once and for all.

[zerocrossing]

…I'm talking about real world use, not sending synthesizer oscillator tones into it…

The sweep used on the Gear Page test was developed by Jay Mitchell, who is an audio engineer, specifically to represent the frequencies of an electric guitar including the upper harmonics, which is where the problem lies, in those complex upper harmonics—not necessarily limited to high frequency fundamentals. Those frequencies would be in use constantly with a chimey electric tone, even on cowboy chords.

As I already said, it seems minor when recorded, but while I play it irks me like a rattling bridge noise, but nothing can settle it. Grating but not earth shattering. Enough to make me jump through a lot of hoops to avoid it though, when more oversampling would settle it once and for all.

Well, if I may be so blunt, Jay Mitchell has is head up his ass, as did I. I pulled it out though, and now I'm fine. Here's why!

I just did a different test, and instead of using a digital source, I used an analog source. I loaded up my Amplitube patch that I used on the lead sound of the track I posted. It's not Tonex, it's Channel 3 of the Dual Rectifier at 7.7 being fed by the Black 76 compressor with a bit of boost. So, not a shrinking violet of gain. Playing as high up as I could and doing a bend, I just plain and simply was not hearing the "Pac Man" aliasing shift. However, the sine from the plugin presents a lot of aliasing (-44 db), but the triangle from the analog synthesizer presents almost none! Zero? No. Not zero, but so low it was hard to measure on my spectrum analyzer. As a control, I took an analog triangle (what a f'n buzzy triangle! ) from the Uno Synth Pro X and ran it through a BK Butler Tube Driver, and did find zero aliasing.

The Dual Rectifier wasn't on my list of problem models, though, so I ran one that was, the Mark IIC+ it's worse, for sure, but I don't get any aliasing at all until I feed it a B5 (987.8 hz) and then it becomes noticeable. I'm thinking it's bad enough to call that one a bug, because the Mark III doesn't act like that at all.

Quick check of the AC Box XV from Guitar Rig 7 that I thought was bad... nothing. False alarm.

Quick check of some high-gain Tonex, generally very good, though faint aliasing was present at very high notes on bright, high gain models. The one I found that was bad was some Orange capture from the user library, and when I played though guitar it, it was painfully bright, even when I added a cabinet IR. Why did I download that?

The moral to this story? Stay in school, kids! Amplitube is clearly just amplifying issues with a digital sound source. I'm sure Jay Mitchell's sound he was using had some problem as well. It definitely sounded weird to me. Take these "tests" with a grain of salt. People post things to justify their own ideologies. There was an Oberheim vs. Diva test here that many got wrong, but the person who posted the blind test was selling a preset pack, so of course they made the plugin sound great and avoided any places where it didn't sounds as good. Diva is a great plugin, but it's no OB-8. I fully admit that I was surprised at my findings, but so surprised that I questioned my methodology. It was flawed, and I fully admit it.

After all that, I still wanted to double check everything in a real world situation for my use case, as I do have a "full analog" amp solution of sorts. My Blackstar HT-5 Dual is a full preamp. It even has a "cabinet out," and let me tell you, that cab modeler sucks ass, so I took the main line out and went into Amplitube as an IR loader for the cabinet. It is true, that using this rig, even fed with the analog synth, gets me zero aliasing at any gain at any frequency. It's not a great sounding preamp, though, but I'm sure there are better solutions for this. Switching to Amplitube and picking some of my favorite amps provided a better sound. A lot better. I didn't buy the Blackstar as a full preamp solution, and as a drive pedal it's good. So, if you're worried about some faint aliasing when shredding on the E string at fret 21 and above, amp modelers are not for you.

If anyone questions this, I wholeheartedly recommend doing the same sort of test. Warning: it's a giant pain in the ass.