Speaker drive and compression - what value is closest to the real thing?

Fixed. Thanks. Though I have to laugh that "u" "mustv" have felt the need to correct my spelling typo.
well, since you mention it, I didn't feel any need at all (being somewhat dislexik I'm not much of a speller so couldn't give 2 sh%ts about spelling perfection). What I was trying to ask was clarification on what you are referring to?
 
Last edited:
So you're saying include a preset selection for speaker types? That could kinda work, but the issue here is that in reality speaker drive would be a function of the amp's volume, the speaker's individual power handling, and the number of speakers. Within any modeler the amp model volume is emulated to some degree but without some significant meta data regarding the IR loaded, how could a modeling device understand the virtual power handling of the IR?

To the later...yes, no speaker is perfectly efficient that's not what I'm suggesting. Have you messed around with the parameter at all? Seriously, I think it's the special sauce at higher settings for those who are looking for the sound of a speaker being pushed past it's limits. But at more subtle settings it seems to add another layer of subtle compression which one may or may not like. I suggest trying it and seeing if it's something you want to use and worry a little less about whether it's 100% "authentic" as I can't see how a single setting here would be accurate in all settings.

There's no need for metadata for the speaker, just like you don't need for the amp like we've talked about.

Yes, I have.

Isn't this the forum from the arguably leading company when it comes to valve amp modeling? Isn't this the company that puts out constant updates in order to get better and better? What I find odd is that a minor criticism, maybe suggesting that something might be wrong - God forbid - is met with "don't worry about accuracy" in this very forum out of all places.
 
There's no need for metadata for the speaker, just like you don't need for the amp like we've talked about.

Yes, I have.

Isn't this the forum from the arguably leading company when it comes to valve amp modeling? Isn't this the company that puts out constant updates in order to get better and better? What I find odd is that a minor criticism, maybe suggesting that something might be wrong - God forbid - is met with "don't worry about accuracy" in this very forum out of all places.

I am not sure people are saying "don't worry about the accuracy" as much as they are saying
the variables are not fixed or static, and that speaker drive and comp depend on a variety of
factors. I am not sure how those context dependent factors could be modeled accurately that
you are asking for. Maybe they can be. I just don't think it is a static/fixed quotient, as much as
a sliding scale that depends on the specific speaker and amp relationship.
 
I am not sure people are saying "don't worry about the accuracy" as much as they are saying
the variables are not fixed or static, and that speaker drive and comp depend on a variety of
factors. I am not sure how those context dependent factors could be modeled accurately that
you are asking for. Maybe they can be. I just don't think it is a static/fixed quotient, as much as
a sliding scale that depends on the specific speaker and amp relationship.


I never asked for a single asbolute value. I'd be ok with a broad guideline "Deluxe Reverb with a vintage Fender oxford you could use 1 with the vol up to 5, past that 2. Plexi with Greenbacks are usually near cranked, so to match that speaker drive should be a tad higher like 4". I'd be fine with something like this for a start. The more detailes the better.

A way cooler idea would the more detailed speaker profile that has been proposed. Add the info about power handling, sensitivity along side the current impedance curve to make the speaker respond to the amp block just (or closer) to how the real speaker would react to any specific amp. That would be ideal IMO.
 
Last edited:
There's no need for metadata for the speaker, just like you don't need for the amp like we've talked about.

Yes, I have.

Isn't this the forum from the arguably leading company when it comes to valve amp modeling? Isn't this the company that puts out constant updates in order to get better and better? What I find odd is that a minor criticism, maybe suggesting that something might be wrong - God forbid - is met with "don't worry about accuracy" in this very forum out of all places.

Maybe I'm being ignorant here but more or less in my mind you might as well be asking what's the most accurate tone stack or gain setting. To me there's zero difference here. Just like one may prefer a cleaner tone or a dirtier tone they may like the sound of a lower wattage speaker pushed or a "clean" high power speaker, that's why there are a range of values. I imagine if you crank up the gain, you may want to up the Speaker Drive as well for "accuracy" and vice versa.

Now if you mean accurate specifically to the Amp specimen being modeled along with the matching IR of it's speaker at specific volumes....more power to you.
 
I tend to gravitate only toward modelling that mimicks real world standards or modded standards - so the standard tone stack controls, tube swaps, bright/fat... switches, changing cap values ... all that's cool to me, even the nerdy stuff. The ones I find challenging are the ones where there's really no real world equivalent that even most nerdy tinkerers can relate to, like speaker drive, speaker compression, Pl Bias, + a number of other advanced ones... these are the ones that I prefer to be automated since I have no way to measure/understand what those do even though I own a few tube amps to experiment with (does not bother me that those controls are there for those enlightened enough to tweak intelligently, I just have no interest in tweaking something I have little or no chance at getting my head around) - by automated, I mean that their values change based on other moving parts to accurately reflect what the average instance of that real amp will do with the readily understandable controls / mods set a certain way. Afaiu, it's pretty well agreed that modelling kind of hits a snag at the cab because either the amp model has no access to info on the real cab connected, or limited info on the IR attached. Having a more defined cab "profile" would seem to enable some untangling of that snag, but it seems this may be another one of those subjects that can only generate interest once Fractal actually implements it - then it'll be considered the best thing since sliced bread to all, which it probably will be if past successes are any indication.
 
Last edited:
Assuming the new speaker drive algorithm from the Axe III makes it's way down to the FM3 and FM9, it's all going to change soon anyway. The new algorithm behaves quite differently from the old one. Generally speaking, lots of speaker distortion just sounds bad and is rarely desirable. The new algorithm covers the lower, more usable range of speaker breakup and is much more of a subtle smoothing and filtering than a crispy fart out at higher settings.
 
I never asked for a single asbolute value.
That’s exactly what you asked for in your first post whether you used those words or not. “…, what should the value be?”, not “What is a range of values” or anything to generalize.
Isn't this the forum from the arguably leading company when it comes to valve amp modeling?
It’s the forum, yes. Say that again 10 times. It’s. The. Forum.

You need to be blocked. Goodbye.
 
Maybe I'm being ignorant here but more or less in my mind you might as well be asking what's the most accurate tone stack or gain setting. To me there's zero difference here. Just like one may prefer a cleaner tone or a dirtier tone they may like the sound of a lower wattage speaker pushed or a "clean" high power speaker, that's why there are a range of values. I imagine if you crank up the gain, you may want to up the Speaker Drive as well for "accuracy" and vice versa.

Now if you mean accurate specifically to the Amp specimen being modeled along with the matching IR of it's speaker at specific volumes....more power to you.
Yes, I want the value for a specific amp set to a specific volume into a specific IR. But that answer might be different values for different volumes. I guess it would be fair to assume that's, but I don't know what the algorithm is actually doing. Maybe it's dependant on the amp block master volume, if that's true maybe the speaker drive is actually modeling speaker age/use/worn. I don't really know.


That’s exactly what you asked for in your first post whether you used those words or not. “…, what should the value be?”, not “What is a range of values” or anything to generalize.

It’s the forum, yes. Say that again 10 times. It’s. The. Forum.

You need to be blocked. Goodbye.

If you read a few posts up I've already said I'd be happy with a range of values if that's the case. Sorry if I wasn't so precise in my question in the original post.

I'm at the right place for asking such questions then. No?

Why is that? Bye.
 
kinda harsh eh
Figure I’ll stick my nose where it probably won’t be appreciated.😉 First, to me my FM3 is for getting a sound that I always wanted to get in a different way without all the extra equipment that I needed to get that sound. I don’t really care if the parameter ranges and specific points match exactly a specific real world piece of hardware. I just tweak until I get there. IMO the critical aspects of building a preset all pretty much default to a nominal useable spot and then can be refined if needed or wanted.

I did have to step back while lurking on this thread and realize there are probably many younger people (or anyone) who have very little experience with some of the amps that are modeled and want to obtain what they hope is an “authentic” recreation. While I have not played many of the amps modeled, I’m old enough that I played enough that it’s just not that important. There are really only so many tube amp circuit topologies and everything else is some variation with some tweaks to make minute changes in the grand scheme.

As far as being “harsh”…yeah I have to agree. What’s that about?
 
Figure I’ll stick my nose where it probably won’t be appreciated.😉 First, to me my FM3 is for getting a sound that I always wanted to get in a different way without all the extra equipment that I needed to get that sound. I don’t really care if the parameter ranges and specific points match exactly a specific real world piece of hardware. I just tweak until I get there. IMO the critical aspects of building a preset all pretty much default to a nominal useable spot and then can be refined if needed or wanted.

I did have to step back while lurking on this thread and realize there are probably many younger people (or anyone) who have very little experience with some of the amps that are modeled and want to obtain what they hope is an “authentic” recreation. While I have not played many of the amps modeled, I’m old enough that I played enough that it’s just not that important. There are really only so many tube amp circuit topologies and everything else is some variation with some tweaks to make minute changes in the grand scheme.

As far as being “harsh”…yeah I have to agree. What’s that about?
Nothing wrong with using ears to dial a sound in for sure - the point I was trying to get at above (lost also I suspect) is that some settings (speaker drive) are not easily determined by the average joe looking for a sound that matches some rig he heard or perhaps owns, and that average joe probably prefers that setting gets approximated automatically (doesn't have to be exact (amp to amp instance differences and all that). I think this is a reasonable wish for these type of parameters and imo is what modelling strives for (to replicate the moving parts of instances of real world amp rigs). The difference wrt one particular automated setting might be very small / nuanced, but added together these bits can add up to a big improvement which is one reason I think Axefx is best of breed (they strive toward ALL the little bits). The amp/cab relationship is one of the more untapped areas of modelling afaik, and one that Fractal has made significant strides into ahead of competition. Trying to silence logical discussion of where the technology might/could go next is, imo, damaging.
 
Nothing wrong with using ears to dial a sound in for sure - the point I was trying to get at above (lost also I suspect) is that some settings (speaker drive) are not easily determined by the average joe looking for a sound that matches some rig he heard or perhaps owns, and that average joe probably prefers that setting gets approximated automatically (doesn't have to be exact (amp to amp instance differences and all that). I think this is a reasonable wish for these type of parameters and imo is what modelling strives for (to replicate the moving parts of instances of real world amp rigs). The difference wrt one particular automated setting might be very small / nuanced, but added together these bits can add up to a big improvement which is one reason I think Axefx is best of breed (they strive toward ALL the little bits). The amp/cab relationship is one of the more untapped areas of modelling afaik, and one that Fractal has made significant strides into ahead of competition. Trying to silence logical discussion of where the technology might/could go next is, imo, damaging.
Yes I certainly am not intending to shutdown dialog. In fact while it is reasonable to take from my comment that I was advocating to just “use your ears”. The truth is I don’t do that absolutely. I tend to sweep the entire range from max to min and some points in between to get an idea what I’m dealing with then use ears and the info I gained doing that, especially with something I’m unfamiliar. For sure if this discussion would lead to some major revelation that Cliff and company haven’t thought of and would makes sense then more power to it.👍
 
Yes I certainly am not intending to shutdown dialog. In fact while it is reasonable to take from my comment that I was advocating to just “use your ears”. The truth is I don’t do that absolutely. I tend to sweep the entire range from max to min and some points in between to get an idea what I’m dealing with then use ears and the info I gained doing that, especially with something I’m unfamiliar. For sure if this discussion would lead to some major revelation that Cliff and company haven’t thought of and would makes sense then more power to it.👍
Apologies - I wasn't clear - my comment about trying to silence such discussions as being damaging was referring to the call above for Tito83 to be blocked - not to your post content. I can appreciate the approach you describe - it's what I do now also - but speaker drive not being a control I've ever seen on a real amp, I can't help but think that as modelling tech moves along more into the amp/cab zone it will be one of those parms that's dynaically set and not a control most users would venture into to tweak though they will understand that it's getting set dynamically according to the makeup of their virtual rig and settings of virtual equivilents to typical real world controls / mods. This is not a criticism of Axefx in any way which is state of the art imo - a state that constantly moves forward.
 
Last edited:
I can see why some people would be concerned about the accuracy of the modeling. As far my experience has shown me, the modeling is very damn good. My FM3 into a Celestion F12 is the best sounding guitar rig I have ever played on. During a period of unemployment in the mid 2000's I was working on tube amps as a side gig. I was also building my own. I had the opportunity to work on, and test, many nice amps, including some vintage Marshalls and Fenders. I can say that my rig is as good or better than anything I have played. It's far better in the sense that it can model so many amps, and speakers. All that versatility in one little box just blows me away.

As for speaker drive and compression, I have tried playing with both. I haven't found that increasing these parameters does anything positive for my presets. I leave them at default values, or try out different settings while running a loop. The normal amp controls have far more impact on tone than the speaker drive and compression. Just like you will dial in your amp tone with drive, bass, mid, treble, presence, master, you name it....you will set it where it sounds the best to your ears. Heavy speaker breakup sounds bad to me, some compression is useful. I get far more response from adjusting the speaker curve and NFB. Maybe the defaults aren't exactly accurate. I have no way of knowing. And to be honest, I don't care. I just tweak until I'm happy with how it sounds. I imagine the new algorithms will work down to the FM3 unless it requires too much CPU. I look at the FM3 as I would any guitar amp. Make it sound like what you want it to sound like.
 
Apologies - I wasn't clear - my comment about trying to silence such discussions as being damaging was referring to the call above for Tito83 to be blocked - not to your post content. I can appreciate the approach you describe - it's what I do now also - but speaker drive not being a control I've ever seen on a real amp, I can't help but think that as modelling tech moves along more into the amp/cab zone it will be one of those parms that's dynaically set and not a control most users would venture into to tweak though they will understand that it's getting set dynamically according to the makeup of their virtual rig and settings of virtual equivilents to typical real world controls / mods. This is not a criticism of Axefx in any way which is state of the art imo - a state that constantly moves forward.
Yes, I understood. No apologies needed. I was just covering the bases because I could see how it might be taken in a more subtle way that I was not exactly endorsing the OP’s concern. It’s all cool. I also decided to jump in because of the “blocking” statement more than anything.👍
 
Seems like you want different values depending on the amp master volume setting. Like one of the first responders said, start twisting knobs and see what your ears like.
 
I can appreciate the approach you describe - it's what I do now also […]
Well-designed interfaces encourage experimentation. One of the top human-interface design guys wrote about the brilliant designs of rotary-dial phones, and door knobs because it only takes a few seconds to explore them and figure out what we need to do to use them.

On a regular amp, we twist the knob and hear how the sound changes. The numbers on the dial are almost inconsequential until later when we want to remember the setting we reached using trial and error and exploring. The interface is real-world physical objects and we're used to manipulating things like that since infancy.

The modeler seems to intimidate people because, while there are knobs to turn, the change often isn't as immediate, and what is changing is probably a lot more esoteric. But, in a wild leap of faith, people can twist the knob… a bunch of times often because encoders act differently than a potentiometer, we hear a change and figure out whether we like that change.

[…] - but speaker drive not being a control I've ever seen on a real amp, I can't help but think that as modelling tech moves along more into the amp/cab zone it will be one of those parms that's dynaically set and not a control most users would venture into to tweak though they will understand that it's getting set dynamically according to the makeup of their virtual rig and settings of virtual equivilents to typical real world controls / mods. This is not a criticism of Axefx in any way which is state of the art imo - a state that constantly moves forward.
I suspect, because modeling is truly in its infancy compared to door knobs and even rotary-dial phones, as Cliff investigates using code to generate the sounds an amp makes, that there'll be a change in how the modeler works, that all the small moving parts we can adjust now will merge into larger units that act more like traditional tube-amp circuitry when we're on the "Authentic" page because things that we need to adjust manually in the modeler would actually be done automatically and dynamically as the circuit compensated and reacted to the change because Cliff will figure out ways to tie them together. I'm sure we'd still have the ability to dive under the cover and adjust individual parameters, but the high-level interface would remove some of the need for twiddling.

It's akin to having a car with an automatic transmission versus a manual; An automatic is usually easier to use, though it has a lot more parts. But once we know how to use a manual transmission and we're more closely coupled to the engine and accelerator, shifting and letting the engine wind up on curving mountain roads becomes a passion because we have more control.

Fractal isn't going to sit still, they're always going to move ahead. Who knows how the interface will change in future hardware and software. Everything about the technology is moving ahead, from the chips to the software tools used to make the interface. Heck, look how far the current generation hardware and firmware have changed in the last few years. I'm amazed when I hit a chord on my guitar and what I hear come out sounds SOOOooo much like my tube amp.
 
Last edited:
I suspect, because modeling is truly in its infancy compared to door knobs and even rotary-dial phones, as Cliff investigates using code to generate the sounds an amp makes, that there'll be a change in how the modeler works, that all the small moving parts we can adjust now will merge into larger units that act more like traditional tube-amp circuitry when we're on the "Authentic" page because things that we need to adjust manually in the modeler would actually be done automatically and dynamically as the circuit compensated and reacted to the change because Cliff will figure out ways to tie them together. I'm sure we'd still have the ability to dive under the cover and adjust individual parameters, but the high-level interface would remove some of the need for twiddling.
This is an important part of modelling imo, and an aspect of Axefx I appreciate a lot: Like no other, it strives to replicate the simplicity of a real amp with just the real life knobs (authentic page), while at the same time providing deep level controls for those that want it (I tend to use some of both depending) and all the while maintaining a high level of authenticity to what happens irl both on the surface and in depth. On the simplicity / authentic mode side I think there is still opportunities for Axefx to adapt more automatically to speaker swaps and cab swaps - Like all aspects of modelling, Fractal is waay ahead of all others wrt adjustability of speaker related characteristics - adding more automation to respond to what people understand more intuitively in this area (speaker swap, cab swap ...) would be the icing on the cake but might require more technical integration with IR makers.


It's akin to having a car with an automatic transmission versus a manual; An automatic is usually easier to use, though it has a lot more parts. But once we know how to use a manual transmission and we're more closely coupled to the engine and accelerator, shifting and letting the engine wind up on curving mountain roads becomes a passion because we have more control.
Though one could think of the authentic page as an "automatic" transmission, some advanced parameters go beyond what I would equate to operating a manual transmission and more into the realm of the weekend mechanic that has their engine apart in the garage to tinker with fuel mixture - anyone can get in there and futz around, but few will know enough to proceed methodically and know what to listen for in order to make adjustments logically and complimentary to other settings. Dynamics page? - ok, woohoo! I got a clutch and shifter!, Plate Bias Excursion? - say whaaaaat? (but by all means, feel free to twirl the Pl Bias Excursion dial "to taste" lol!).
 
Last edited:
Back
Top Bottom