Can tube sound be replicated via plugins?
Jun 26, 2017 at 11:30 PM Post #46 of 179

100VoltTube

100+ Head-Fier
Joined
Jul 29, 2016
Posts
229
Likes
39
I know within the guitar amp world, some solid state amps with software suites have been able to perfectly mimic tube amplification sound and pass double blind tests. So why not with a headphone amp?
Guitar amps are designed to produce massive amounts of distortion, and their speakers are terrible from a hi-fi standpoint, meaning that the DSP has a lot of room to add IMD and all sorts of nasty stuff without the listener being able to hear it.

A high fidelity amp should be transparent, ie amplify the signal without adding or detracting anything else to the signal. A well designed SS or tube amp are both capable to achieving this without any colour or perceptible distortion, providing it is operating within its output and headroom parameters.
Very true, but there are a lot of exceptions to the rest of that statement, except that tube amps are higher maintenance (except nuvistors, but good luck finding an amp that uses them)

Carver proved in the 1980s that the sound of any amp can be copied
He did this for only 2 amps. Mind you, this was 30 years ago, so the quality of the modeling has improved, but so has the quality of the amps being emulated. The signal nulling is just a way to quickly figure out the differences between two amps, not a way to emulate an amp.
 
Jun 27, 2017 at 1:44 AM Post #47 of 179

pinnahertz

Headphoneus Supremus
Joined
Mar 11, 2016
Posts
2,072
Likes
738
Very true, but there are a lot of exceptions to the rest of that statement, except that tube amps are higher maintenance (except nuvistors, but good luck finding an amp that uses them)
David Manley would disagree that tube amps are higher maintenance...or that they should be. In "The VTL Book" he goes on about how proper operation of a tube results in a very, very long life. He also talks about making tube amps transparent.
He did this for only 2 amps. Mind you, this was 30 years ago, so the quality of the modeling has improved, but so has the quality of the amps being emulated. The signal nulling is just a way to quickly figure out the differences between two amps, not a way to emulate an amp.
Actually, the null method is hyper sensitive to differences, and a rather difficult way to match amps. But when you get it, it's a very, very close match. I'm not sure how you'd differentiate between eliminating differences and emulating.
 
Jun 27, 2017 at 9:02 AM Post #48 of 179

gregorio

Headphoneus Supremus
Joined
Feb 14, 2008
Posts
4,158
Likes
2,308
Guitar amps are designed to produce massive amounts of distortion, and their speakers are terrible from a hi-fi standpoint, meaning that the DSP has a lot of room to add IMD and all sorts of nasty stuff without the listener being able to hear it.

You've got that kind of backwards! A DSP is not an analogue device, it doesn't add noise, IMD, any other distortion or anything by itself. A plugin which is perfectly linear does not require any programming at all! On the other hand, adding IMD and any sort of "nasty stuff", whether the listener can hear it or not, requires programming and is therefore more difficult than not having IMD and/or "nasty stuff". If anything, a guitar amp is more difficult to program because there can be relatively little or massive amounts of distortion, typically the response is highly non-linear and, that doesn't mean you can get away with any old "nasty stuff" because guitarists are often intimately familiar with the PRECISE nastiness, distortion and non-linear response of the amps being modeled/emulated.

Castleofargh and a previous poster had it right but to elaborate a little: Tube guitar amps were modeled at least 15 years ago and about 20 years ago some of the most prized tube based mastering compressors/limiters were modeled. Admittedly, the programming complexity and DSP capabilities weren't there to begin with and although you could just about get away with them in a bind, they were otherwise avoided by the serious pros. However, that's not the case today. The vast majority of the old school die-hard mastering engineers, the most critical of listeners, eventually shelved their prized vintage gear a few years ago and got the plugins. An interesting case is Avid's Eleven Rack, originally a plugin emulation of various vintage guitar amps (and cabs), it became so popular with guitarists in the studio that they designed a hardware unit which emulates the plugin, for live performance use by guitarists!

So yes, it is entirely possible for the "tube sound" to be emulated by plugins and in some cases has been for years. It takes years of development to get to the point that even the pros can't hear a difference and the problem is that there just isn't the financial incentive to do this for consumer tube amps. The audiophile market is a relatively small niche market, only a small percentage of that already small niche market have any interest in that "tube sound" and only a small percentage of those would be interested in a plugin emulation! So although it's entirely possible to create an essentially perfect plugin emulation of a consumer tube/amp, I'd be surprised if there are any good ones, I can't see how it would be financially practical to invest the required time and money.

G
 
Last edited:
Jun 27, 2017 at 11:25 AM Post #49 of 179

bigshot

Headphoneus Supremus
Joined
Nov 16, 2004
Posts
23,307
Likes
4,865
Location
Forests of the Pacific Northwest
My guess is that even if they built a perfect tube emulation DSP into a consumer headphone amp, audiophiles would refuse to use it because it's the glowing tubes and aesthetic qualities of tube amps are what they really want, not the sound quality. You could design a DSP that would be identical in double blind testing and they still would believe that the tube amp sounded better.
 
Jun 27, 2017 at 1:17 PM Post #50 of 179

pinnahertz

Headphoneus Supremus
Joined
Mar 11, 2016
Posts
2,072
Likes
738
My guess is that even if they built a perfect tube emulation DSP into a consumer headphone amp, audiophiles would refuse to use it because it's the glowing tubes and aesthetic qualities of tube amps are what they really want, not the sound quality. You could design a DSP that would be identical in double blind testing and they still would believe that the tube amp sounded better.
Just add a couple of tube sockets with filament voltage on them. You'll have the heat, and the glow.

The problem I have is using the words "perfect" and "tube sound" together. Tube sound is horrifically flawed, and because there are so many variables, what does anyone call the "perfect tube sound"? It's like the analog tape emulator plugins. Why would anyone want to do that to perfectly good audio in the first place?
 
Jun 27, 2017 at 1:36 PM Post #51 of 179

bigshot

Headphoneus Supremus
Joined
Nov 16, 2004
Posts
23,307
Likes
4,865
Location
Forests of the Pacific Northwest
By perfect, I mean a digital real time filter that exactly duplicates the sound of a tube amp. Someone earlier said it would be impossible to create a DSP that sounds like tubes. Gregorio is absolutely right that there are pro filters that do it. Why not build that into a consumer headphone amp? My guess is that sound isn't the reason audiophiles like tubes
 
Jun 27, 2017 at 2:59 PM Post #52 of 179

100VoltTube

100+ Head-Fier
Joined
Jul 29, 2016
Posts
229
Likes
39
@gregorio: what I meant was that the DSP modeling in guitar amps can add extra nastiness that doesn't exist in the target amp. In the pro audio applications, it can't, so I guess they do have some good DSP for it there.

What I'm trying to get at is that even though stereotypical "tube sound" (high levels of 2nd order harmonic distortion...) can easily be produced by plugins, you can't make a FET or a BJT into a triode, or offer the low input capacitance of any tube. Even a massive, 70 year old transmitting tube (like a 211 or a 803) has a lower input capacitance than a teeny tiny JFET. Triodes also have a finite plate resistance, which can be used to the advantage of the triode, and some offer exceedingly linear performance sans NFB, which no readily available transistor that I'm aware of can do. Tubes are not inherently objectively worse than transistors, and when used correctly, they offer advantages. Transistors do have other advantages, like high transconductance, and theoretically infinite source/collector resistance. Transistors and tubes are always going to be different from an electrical standpoint, which cannot be changed by DSP. Neither technology is "superior;" they are different, and have different uses.

I do agree that the audiophile crowd that likes the stereotypical tube sound will never be sold on DSP in place of their glass, even if it can't be distinguished from the actual tube amp in blind testing.

In response to the question asked by the OP: Yes you can replicate "tube sound", but it helps to put a resistor in series with the output, unless it's a preamp. However, you can't turn a cheap and schitty SS amp into a masterfully designed amp (hybrid or all tube) that uses tubes to its advantage.
 
Jun 28, 2017 at 6:38 AM Post #53 of 179

gregorio

Headphoneus Supremus
Joined
Feb 14, 2008
Posts
4,158
Likes
2,308
It's like the analog tape emulator plugins. Why would anyone want to do that to perfectly good audio in the first place?

There's three reasons and I have applied tape emulation plugins for all of them: 1. The client is misguided and insists on an "analogue tape sound". I even had one client who wanted the analogue tape sound but wanted it to sound much cleaner than analogue tape. Obviously that's a contradiction but I gave him what he wanted by just adding certain tape emulation parameters, something I could only do with a plugin and not with actual analogue tape. 2. In film/TV I've been given a digital master of music to include but the scene shows the music being played on a tape machine. I've done this with vinyl emulation plugin too. 3. Contrary to much audiophile belief, music production is not necessarily about fidelity, it's about art and what sounds subjectively better and that can mean less fidelity, at least in places or for certain instruments/sounds. Personally I rarely use tape emulation for this reason but I have done on occasion.
In response to the question asked by the OP: Yes you can replicate "tube sound", but it helps to put a resistor in series with the output, unless it's a preamp. However, you can't turn a cheap and schitty SS amp into a masterfully designed amp (hybrid or all tube) that uses tubes to its advantage.

It's not just about 2nd order harmonic distortion, you can emulate anything, including resistors. The Eleven Rack I linked to previously doesn't just emulate guitar amps, it also emulates various vintage guitar cabs and the mics typically used to mic guitar cabs and, each of these elements is interchangeable. The idea of the unit is that you can plug your guitar into the input and take the output direct to the mixing desk, no amps, speakers or mics required:

"Digidesign went to great lengths to model essentially every component in each amp. “It was an incredibly painstaking process. We created models for the tubes, tone stack, power supply, preamp, power amp—you name it.” ... By modeling each of those components, Eleven provides players with a much more authentic re-creation of the original amp control set than other amp simulators. “Diehard tone seekers who want to experience the VOX AC30 or Marshall Super Lead will be able to dial up the same settings on Eleven as they would on the real amp,”. The Digidesign team also emulated aspects of amps and cabinets that are frequently overlooked in other modeling units, like power amp sag and speaker cone breakup. ... " -
http://akmedia.digidesign.com/news/docs/Behind_Eleven_42323.pdf

Not that I'm a fan of quoting marketing materials here but it does show that pretty much any behaviour or component (including individual tubes) can be emulated with a level of precision that even engineers and the guitarists themselves are satisfied with. Again though, in the case of consumer tube amps there would be no point, there's not enough of a market to make properly developing such a plugin financially viable.

G
 
Jun 28, 2017 at 12:56 PM Post #54 of 179

frodeni

100+ Head-Fier
Joined
Aug 27, 2015
Posts
141
Likes
27
Many say tube sound is nothing more than EQ and designer coloration, if this is the case would it be possible to exactly replicate tube sound through the use of a plugin?

Lets stay with current state of tech and common understanding of the topic.

Those who typically tend to believe in measurements and typically has a technical view on sound reproduction, argue that the sound made by tubes, is distorted. This basically stems from the fact, that this is what is measured. Using the positivists view, this is usually proven with amplitude charts, and some rare instances of more complex waveform tests. It is not really understood why these measurements do not correlate with the listening experience people report of having. If you move along this paradigm, it should be possible to make a filter that distorts the sound, that would reproduce the tube effect.

The main other school of thought, are those who trust their hearing, rather than the measurements. People in this category, in which I tend to be, do not get the same experience with any of these tube filters. It is simply not what our ears tells us.

Let me just add, that I have no idea why there is a huge difference in what the theory tells me should be the case, and what my experience is. The theory about the measurements appear sound as, well, something. Yet my experience is in something completely different.

My guess is that sometime in the future, this will be better understood. When that lack of understanding is bridged, we probably should be able to produce a single piece of gear, that would be able to reproduce current solid state and tube rendering. My guess is that by then, both renderings would need some form of filter, as the current rendering of any gear, simply is not perfect.

The main thing about tube rendering, is the speed. The perceived speed. It is also more raw, as in a bit lacking in precision. This gives it a raw feel, and tend to add some life to the music. It is a loved rendering for guitars. The randomness is nor easily generated by any equalizer, as there is no randomness to it. Also, the ground is actually floating in many tube designs, which tend to add effects that are complex. My guess is that those just thinking this is a amplitude thing, hits the mark as lack of technical understanding of the innards of the tube design. Also, many tube design are sensitive to interference. In my experience, extremely sensitive. Designing a filter to mimic that, would require deep insight into the innards and functionality of tube designs. It probably is nothing but simple. The truth is that all this is just speculation, as it would be for anyone else, as we do not understand this, and certainly do not know.

What you probably could do, with relatively ease, is mimic the amplitude curve of a know tube amplifier. There probably are some other tricks that seem to make sense as well, I just have no detailed insight into the current state of the matter. However, you would find a lot of people who claim that such filters do not sound like the tube rendering it is supposed to mimic.

In the photographic world, there is something called and "unsharpen mask filter": It actually reduces sharpness, but for us humans, the image appear sharper. My guess is that there is something like this going on, just that in the case of audio, this is simply not understood yet. You would still find people claiming that a signal converted back to analogue using a DAC, then converted to a vinyl record, then read with a pickup, fed through a RIAA amp, then fed into a system, actually improves the sound quality. Even if you use the exact same DAC, and feed it directly into the amp, skipping the vinyl loop. Why all those vinyl stages should contribute to any accuracy, is really hard to grasp, yet people seem to have just that experience. These are the same people who tend to love tube amps. Just more of the same thing.

Both camps are probably right, given how they argue. There is just a bridge lacking. A bridge of understanding.
 
Jun 28, 2017 at 3:16 PM Post #55 of 179

danadam

Head-Fier
Joined
Jan 2, 2017
Posts
93
Likes
62
In the photographic world, there is something called and "unsharpen mask filter": It actually reduces sharpness, but for us humans, the image appear sharper.
Is that correct? The mask, yes, has reduced sharpness, but that mask is then combined with the original image. Can you really say that objectively the resulting image has reduced sharpness comparing to the original and only appears sharper to us?
 
Jun 28, 2017 at 4:27 PM Post #56 of 179

castleofargh

Sound Science Forum Moderator
Joined
Jul 2, 2011
Posts
9,523
Likes
4,919
reproduction is a matter of objective fidelity. we measure the copy and compare to the reference. we don't need somebody to hold a kilogram in their hand and tell us how good the weight is at being a kilogram.
I don't ask a measurement rig if the music feels good, and I don't ask for human impressions to tell me about objective fidelity. it's that simple. we just ask a specific question, actually try to answer it, and most of the time the ambiguity is gone. here it's about making a digital copy of the signal coming out from a tube amp. can we do that well? most of what we know seems to move toward: yes we can.
now will someone be able to feel like its the tube amp sound if he doesn't see the pretty glow and isn't lied to about the source of the sound? I'd venture that most of the time the answer could be no. it would have little to do with the objective fidelity of the copy and a lot to do with cognitive bias.
 
Jun 28, 2017 at 5:13 PM Post #57 of 179

100VoltTube

100+ Head-Fier
Joined
Jul 29, 2016
Posts
229
Likes
39
you can emulate anything, including resistors
Without knowing the impedance curve of the transducer, you cannot model the FR chance in response to the output impedance, and you cannot accurately model the change in electrical damping on the impulse of the driver without the Q-factors of the drivers. Good luck getting the manufacturer to divulge the thiele small parameters of their drivers.

There are a lot of different ways to measure amps that are not ordinarily used, which can reveal flaws in otherwise well-measuring pieces of equipment, like opamps, as documented in this paper: http://www.nanovolt.ch/resources/ic_opamps/pdf/opamp_distortion.pdf
But at a point, I'm not sure whether these flaws matter if they can't be detected in blind testing. Same goes for the differences in emulated tube amps and real tube amps. If you can't tell the difference, then for all intents and purposes, there is none.
 
Jun 30, 2017 at 4:14 AM Post #58 of 179

frodeni

100+ Head-Fier
Joined
Aug 27, 2015
Posts
141
Likes
27
Is that correct? The mask, yes, has reduced sharpness, but that mask is then combined with the original image. Can you really say that objectively the resulting image has reduced sharpness comparing to the original and only appears sharper to us?

USM is a destructive algorithm. At least it used to be, as of the USM filter we used in the signal processing class at UiO. Not really my field, and this was at introductory level of signal processing. It consistently and always described as an destructive algorithm.

I saw this video on Youtube claiming that EQ algorithms are non destructive. If inverse value is applied, nothing is lost. Again, not really my field.

reproduction is a matter of objective fidelity. we measure the copy and compare to the reference. we don't need somebody to hold a kilogram in their hand and tell us how good the weight is at being a kilogram.

I don't ask a measurement rig if the music feels good, and I don't ask for human impressions to tell me about objective fidelity. it's that simple. we just ask a specific question, actually try to answer it, and most of the time the ambiguity is gone. here it's about making a digital copy of the signal coming out from a tube amp. can we do that well? most of what we know seems to move toward: yes we can.

now will someone be able to feel like its the tube amp sound if he doesn't see the pretty glow and isn't lied to about the source of the sound? I'd venture that most of the time the answer could be no. it would have little to do with the objective fidelity of the copy and a lot to do with cognitive bias.

The trick is to get people to find a distinct sonic trait that is significant enough to reliably tell the difference. Just feeling the flow, or something like that, well, it might be something. And there might be green men on Mars. If a distinct and significant sonic trait is present, and the listener is able to clearly hear it, as in separating gear, a blind test should not make any difference or not. A lot of the traits are just barely audio able, and even if there, cannot be reliable distinguished.

If you hand me a paper with Chinese writing, I cannot tell you what it says. It still may say a lot. If you run a blind test on people who do not know Chinese, and tell them to separate Chinese speech, and they cannot, does not mean there is no difference or meaning in what being said. A true positive must be possible.

As for this kilogram of yours, it is called mass, not weight. You are speaking of the gravitational force on the kilogram, which is dependent on the force, while the mass is constant. This kind of argumentation only works for things stable through time. The experience of love, or music, is not. It simply is not universal. Just like language drifts by time. In a 1000 years, the meaning of the language of today, will not equal that in a 1000 years. Testing physics laws, are expected to harvest the same results in a 1000 years, if conducted exactly the same way.

There is limits to the objective perspective and limits to the subjective perspective.

It is hard to peel an orange with a sledge hammer.
 
Jun 30, 2017 at 7:43 AM Post #59 of 179

castleofargh

Sound Science Forum Moderator
Joined
Jul 2, 2011
Posts
9,523
Likes
4,919
weight vs mass. when somebody tells you he/she has lost weight, you ask for the value in newton and everybody thanks you for bringing the much needed correction? you talk about language, when something is misused by almost everybody, doesn't it become the norm in daily language? anyway, you're right and I'm not. I should even have expected it coming from you so it's a double fault for me.

about the rest, I define audible difference by how I can or cannot pass a blind test for it. call that arbitrary, oversimplified, or flawed if you will, but it offers me a clear ranking of how noticeable something is to me. so I never have to pretend like something is important when I can hardly ever notice it. you know, like jitter at -120dB, ringing at 20khz, etc.
and of course I stand by what I said about making a copy of tube sound. to measure actual fidelity, measurements are kind of mandatory and much more accurate than asking for people's opinion. if we make a lousy copy and wonder if the variations are audible, then sure we test people.
 
Jul 10, 2017 at 7:29 PM Post #60 of 179

Orestes1984

100+ Head-Fier
Joined
Jul 29, 2016
Posts
179
Likes
34
A "tube" sound is distorting the audio in some way, not improving the accuracy of reproduction.

Really.... actually I don't realise this at all. Just like solid state digital amplifiers there are tube amplifiers that distort the sound quality in some way and there are those that don't. This statement is an utter load of nonsense. You might prefer solid state digital amplifiers over tube/valve amplifiers but again this does not make the statement held above true.

Either way if you design a solid state amplifier and its topology correctly and a tube amp and its topology correctly both systems will be capable of achieving sound without any colouration and without an perceptible distortion all things considered and provided its operating within its capabilities and headroom. The idea that valves distort comes from the wanted harmonic distortion from musical instrument amplifiers. This is not the reality of all valve amps. Some people just like to make sweeping generalisations without ever having used a valve amplifier in their life.

If one is to make a proper comparison of valve vs. digital one must use both systems first. Lots of people like to say things without trying something first. Despite popular opinion neither option is expensive or wrong. There are a lot of vintage valve amplifiers one can have for less than $1000 even less than $500 that will stack up well even today. Price comparison is not the issue here either.


If we're going to talk about just my personal setup, not preferences or whatever, I have a pentode amplifier in push pull configuration using four 6BM8 and three 12AX7s it sounds nothing like the sweeping generalisations most people have of tube amps and to be fair. I listen to my amp for music reproduction not music distortion. I've heard both, I've got a Marshall JMP stack sitting in my dads living room the thing is, that has a purpose. My valve amp has a completely different purpose.

We're not listening to music to be rock stars here (or at least I'm not) so I really wish the cliche of harmonic distortion and "warmness" would go away.
 
Last edited:

Users who are viewing this thread

Top