Turning a traditional guitar cab into a 'flat response' cab via digital EQ

[kevinerror]

I wanted to start a discussion about a technique I thought of about a month back that I've had fairly positive results with. I have a normal, traditional, non 'FRFR' guitar cab, and I've been able to flatten the frequency response/output of the speakers in the cab with various digital EQ techniques. This way, I can use cab IRs and emulation into this cab as if it were a 'FRFR' type setup (though not actually 'full range'). I thought I'd share my findings with you all and see what you think!

First off, there are certain limitations that I am well aware of, and I know I cannot get around. Let me state those from the getgo so that we're all aware of the context of what can and can't be done:

1) A single sized driver cannot produce a full-range (in this context, full range meaning 20z to 20khz) of frequencies. This is just a fact. I consider this to be acceptable, as guitar doesn't take up all of the space in the 'full-range' spectrum. I am personally of the opinion that the highest frequency that any 12 inch driver can produce is more than high (in frequency) enough for the sound of a conventional guitar signal.

2) This process involves reference mic measurements/placements, so there are some natural variances not being accounted for: mic position, the 'flatness' of the reference mic used, differences in frequency response at different amplitudes, temperature, room reflections, etc - I am personally also of the opinion that these variances, while certainly existent or perhaps even measurable to some degree, are too small to matter in any practical application.

3) A lot of the gear mentioned in this post is very, very cheap - the opposite of the Axe. I know, I get it - why would I pair up a $2k unit with such cheap stuff? I tried not to think of it that way, and I was pleasantly surprised by the performance of all the pieces used.

With those disclaimers out of the way, here is what I've found.

I have an older 300 watt fender 4x12 guitar cabinet that I wanted to test this theory on. The speakers are some kind of stock fender something-or-other - frankly they sound like crap. But, they handle a decent amount of power, and they are fairly lightweight, so this cab was a nice fit for me. My theory was that guitar speakers have a non-linear frequency response that could most likely be EQ'd out to be somewhat flat - the same way that a recording of a guitar amp (or even vocals, drums, etc) can be EQ'd so that sharp peaks and valleys can be corrected, if desired. This is just a more extreme version of that technique. This would yield a cabinet that has minimal coloring on the signal it's given - This is PERFECT for modeling setups, as I don't want to be stuck with the (crappy) response of my cab - I want to be able to change it up and use whatever IRs I'd like, just like a FRFR setup. So - making my guitar cab as flat as I could was the name of the game.

I started trying to do this process 'manually' using parametric EQs and various RTAs/spectrum analyzers along with a flat-response reference mic, using white/pink noise and comparing the difference between what I sent to the speaker and what came out. I actually was able to get it to be nearly perfect (meaning, nearly the same between input and output), but it involved something somewhat ridiculous like 15 specific bands on a software parametric EQ (Fabfilter Pro-Q for those curious), and that was going to be tough to recreate accurately in the Axe (or any other modeler, really). It could be done, but every EQ is a little different, and it probably wouldn't be as perfect.

Instead, I found a rack mount processor that, among other things, has a 31 band +/- 15db graphic EQ with an 'auto EQ' function. The main piece of gear to do this was a Behringer (I know, I know..) DEQ2496.It basically automates the process that I was doing manually. It sends out pink noise, measures the 'response' of whatever you've got your reference mic pointed at, and attempts to flatten the frequency response of the system by adjusting it's built in 31 band graphic EQ. It's original design was intended for tuning rooms and sound systems, but the process/principle is the same in this context - it's flattening the response it receives to be as close to what its sending out as it can.

So, I used this unit to send pink noise through my cabinet, and used my flat-response reference mic (It was from Dale Pro Audio) to take measurements. After trying a few different positions and seeing how the unit EQs the response, I came up with a setup that was really very good. Sound coming out of the cab was very, VERY neutral. It was super easy to dial in a tone with headphones, then plug that tone right into the cab through the processor, and have it sound the same. I was super happy with this!

I took it one step further - I didn't want to use this piece of gear at shows (It's notoriously unreliable - very good at what it does, but unreliable), so I captured an IR of the EQ settings it setup with the Axe. I just hooked it up the same way you'd hook up a mic to a cab to capture an impulse - Out of the Axe, into the unit, out of the unit, into the Axe. Then, in my patches, I split my signal at the very end, and sent it through this new IR (Yes, In series - remember, this is done to flatten the response of the cab) and to Output 2, then to my cab. I got the same awesome flat result. This means that I can essentially turn my cab into any other cab however I see fit - Just like a FRFR setup. I could even take an IR of the original EQ curve I made in software, and get the same results.

So I'm way happy with this - It works wonders. I can't really 'prove' this to anyone who isn't in the room hearing it, but I can tell you that this is a very legitimate and effective way to utilize IRs and modeling. I've jammed with it and played out with it, and it's no different in feeling than when I was using a PA speaker as my guitar amp. I consider this to be a success.

However, everyone I've mentioned this to just writes it off on principle. Here are the issues people seem to have:

- Yes, I know.. two IRs in series, one after another.. You have to consider that the 2nd IR isn't emulating a cab, it's emulating a very specific EQ curve that is specific to the loudspeakers I happen to be using. It's an odd use of the Cab block, but it does exactly what it's supposed to.

- 'Behringer? Dale Pro Audio? Not near my Axe!!' - People need to get over this. They do what they're supposed to do very well. I know cheap units can be unreliable, but context is critical here In this setup, the units were used for a total of about 10 minutes, and the results of their use was saved in an IR, so their reliability isn't really an issue. I'm probably selling the DEQ unit soon - I don't really need it anymore!

- 'Why not just get a FRFR setup?' - Yeah, that might make more sense for some, but I wanted the aesthetic of the guitar cab, I already had, and I can't afford 4 12 inch speakers with a flat response right now (they probably won't even be that flat). This seemed like a method that made more sense for my situation, and it's turned out great.

I'd welcome any discussion on this - I predict a lot of people will again 'write this off' totally because it's not the way things are normally done.. but I'd love to see what people think Thanks for reading!

 

[kevinerror]

I just realized that this perhaps could be done with the Tone Match block... match the sound of Pink noise through a cab to actual pink noise.. That would probably work exactly the same, if not better - I don't know how many 'bands' or even what method the Tone Match block uses... I'll have to investigate!

 

[jimfist]

fvck 'em if it works for ya

 

[Morphosis]

The main piece of gear to do this was a Behringer (I know, I know..) DEQ2496.(...) so I captured an IR of the EQ settings it setup with the Axe.

I´ve done this similar: Not for traditional cab, but for my frfr. Instead of using DEQ2496, i use IK Multimedia ARC and shoot from that frequency correction result an IR. Works.

I just realized that this perhaps could be done with the Tone Match block... match the sound of Pink noise through a cab to actual pink noise.. That would probably work exactly the same, if not better - I don't know how many 'bands' or even what method the Tone Match block uses... I'll have to investigate!

Back to traditional Cab correction with TMA-Block (instead of ARC or DEQ2496 or others...):

TMA match from 20Hz - 20kHz. A average guitar cab will have a response between ... hmmm .... round 80Hz - 5kHz. That said, TMA will try to compensate frequencies, the speaker can`t produce (with acceptable energy input). So a TMA correction will boost under round 80Hz and over 5kHz that much energy, that you loss all the headroom...

To prevent that, let the reference signal not show these not reproducable frequency-area.

How to do:

Synth Block as measurement signal:

a) Pink noise:
Take off voice 1 & 2. On voice 3 set track "off" and select "pink noise". Turn up frequency to 20kHz.

b) sine sweep:
Instead of "pink noise" select "sine", parameter "frequency" connect via modifier with lfo. lower the rate for better results

To prevent from measuring uncorrectable frequency areas:
Place two PEQs after the synth block and cut by blocking with strongest slew rate under round 80Hz (low-cut) and above 5kHz (hi-cut) the measurment signal. Twice Peqs in seriell will make the slew rate stronger....

Send this shaped signal through the cab and via measurement mic back to the Axe-Fx (tma=local).
Same synth->2x PEQ signal direct to the tma block (=reference)

This could be a workround with tma block, to do the DEQ2496 job (with much higher resolution!).

settings TMA block: "live" (Dual FFT behaviour)

with pink noise: average time default 3.5 or higher (5.0)
with sine sweep: average time = Peak hold (sweep as long, as both measurment curves completely frozen).

Good luck

 

[fractalz]

I've had a similar idea and thought of doing an inverse TMA of the cab and then serial combining it in cab lab with the IRs I want to use.

So, this would be shaping one guitar cab to sound like another and shouldn't require 20-20k reproduction.


Sent from my iPhone using Tapatalk

 

[Ant Music]

I'm keen to hear how this all works out in practice. I'm surprised there hasn't been more talk about this kinda thing.

 

[RobinMatrix]

I have been playing with using inverse FIR's to correct speaker responses for a while, and there is some software around that can do it for you, however most of it is Linux command line stuff so it's not particularly user friendly, unless you are happy with an xterm as your user interface. There is some windows stuff about, but it seems to crash more often than it runs ... and I've not been overly impressed with the results.

If there is interest, I will look at producing inverse FIR's for our NL series stuff. These speakers have fairly gentle and extended frequency responses, so are a good candidate, and I have some stuff I worked on already one or two upcoming projects. If I produce a generic one, it will be done half-space in an open area, so it should come out reasonably flat when played in your room, then you can add another IR for the cab.

Remember though that a guitar cab produces small, but significant, levels of harmonic distortion, usually more than you get from a true FRFR speaker with its "PA" style cone, so it won't get rid of that, and I usually limit the maximum corrections to around 6dB. I might need to read up a bit on how to produce the output in exactly the correct format to load back into AxeFX, as all the stuff I have to date is just in cell coefficient format for dumping directly into the coefficients of a DSP. The smoother and flatter and lower distortion of the speaker you start off with, generally the better the overall result. I found it was generally an improvement on just using the bare speaker, but still not as good as a real FRFR setup.

Let me know if there is any interest in that, its a few hours work, but it can be done. In theory, you could use Fractal Cab Lab to combine the "inverse FIR" of the real cab with your IR that you want to use, doing it in a single IR block, effectively pre-compensating some of your IR's for use with a traditional speaker.

 

[Morphosis]

I have been playing with using inverse FIR's to correct speaker responses for a while, and there is some software around that can do it for you, however most of it is Linux command line stuff so it's not particularly user friendly, unless you are happy with an xterm as your user interface. There is some windows stuff about, but it seems to crash more often than it runs ... and I've not been overly impressed with the results.

If there is interest, I will look at producing inverse FIR's for our NL series stuff. These speakers have fairly gentle and extended frequency responses, so are a good candidate, and I have some stuff I worked on already one or two upcoming projects. If I produce a generic one, it will be done half-space in an open area, so it should come out reasonably flat when played in your room, then you can add another IR for the cab.

Remember though that a guitar cab produces small, but significant, levels of harmonic distortion, usually more than you get from a true FRFR speaker with its "PA" style cone, so it won't get rid of that, and I usually limit the maximum corrections to around 6dB. I might need to read up a bit on how to produce the output in exactly the correct format to load back into AxeFX, as all the stuff I have to date is just in cell coefficient format for dumping directly into the coefficients of a DSP. The smoother and flatter and lower distortion of the speaker you start off with, generally the better the overall result. I found it was generally an improvement on just using the bare speaker, but still not as good as a real FRFR setup.

Let me know if there is any interest in that, its a few hours work, but it can be done. In theory, you could use Fractal Cab Lab to combine the "inverse FIR" of the real cab with your IR that you want to use, doing it in a single IR block, effectively pre-compensating some of your IR's for use with a traditional speaker.

I`ve done it with RoomEq Wizard ( www.roomeqwizard.com ). Load in a .wav IR, do the inversion and save it back as .wav . Convert with Cab-Lab to UR. Done

 

[t0aj15]

Sounds like someone trying to make a silk purse out of a sow's ear to me. :roll

Guitar speakers and cabs are to narrow in their frequency response to ever even approach full flat response no matter how much EQing is applied. No amount of EQing is going to add frequencies that the speaker/cab combination just is not capable of. If it could be done, it would've been done a long time ago.

 

[Morphosis]

Guitar speakers and cabs are to narrow in their frequency response to ever even approach full flat response no matter how much EQing is applied. No amount of EQing is going to add frequencies that the speaker/cab combination just is not capable of. If it could be done, it would've been done a long time ago.

Nobody said anything else. Bu perhaps it is worth a try to compensate the frequency-range, a guitar speaker si able to reproduce, because when using with IRs: Those will represent a similiar frequency range ... thats the idea ...

 

[luke]

While the OP is super defensive, the reality, when all digital, the end the result, not the means, is all that matters. In each block there are EQ limitations and sound signatures, then the signal goes to the next block and is altered more, often in contradictory ways. Your 1k may have been up, down and sideways six different ways before the finally reading of it.

FWIW, a corrective (read: inverse) IR was all the range when the Verve 12MA FRFR was de jour in Axe world.

 

[RobinMatrix]

Guitar speakers and cabs are to narrow in their frequency response to ever even approach full flat response no matter how much EQing is applied. No amount of EQing is going to add frequencies that the speaker/cab combination just is not capable of. If it could be done, it would've been done a long time ago.

You are correct, however, it is also worth remembering that you are at the end of the day only trying to reproduce frequencies and levels that originally came out of a guitar type speaker in the first place, so the argument that they are "too narrow in their frequency response" doesn't make any sense to me. They are perfectly suited ... the trick is to start out with a cabinet/speaker that has AT LEAST the bandwidth as the one you are trying to emulate and a reasonably smooth response curve, and you can get reasonably close.

The NL series cabs are flatter and have a more extended frequency response than many guitar cabs, so they can be made flat with an inverse FIR to at least 7.5KHz, which will allow them to reproduce with some accuracy a large part of the spectrum.

I'll try and process a few NL series cabs this week and publish the inverse FIR's in 3 or 4 correction strengths, it's certainly not FRFR, but it does correct the cab a little and can neutralise their character somewhat allowing you to overlay another IR,

 

[fractalz]

NL212 would be appreciated.


Sent from my iPhone using Tapatalk

 

[kevinerror]

Sounds like someone trying to make a silk purse out of a sow's ear to me. :roll

Guitar speakers and cabs are to narrow in their frequency response to ever even approach full flat response no matter how much EQing is applied. No amount of EQing is going to add frequencies that the speaker/cab combination just is not capable of.

This is why I didn't use the phrase 'full range' at all - I am not claiming to add frequencies that couldn't previously be reproduced, I can't beat physics. I'm attempting to shape the curve of the frequencies it CAN produce.

If it could be done, it would've been done a long time ago.

This is actually done a lot and has been for some time - it's done all the time at live venues - graphic EQs are often calibrated to flatten out (or whatever the desired term or goal is) the response of a sound system.

As much as everyone would like to think a guitar cab is somehow magically different, in principle, it's not. It's tailored to have a specific range and response that is vastly different from a flat system, but the theory and application is still the same.

I just experimented with more stuff last night using the Tone Match block. Limited all shaping between around 60hz and 10k - what an improvement! Seriously great results. Tried noise, music, guitar, all kinds of stuff. After some minimal additional EQ to remove a little bit of rumble, it all sounded great. Just about the same as my PA speaker (though obviously missing some high end, which I don't need). RobinMatrix has a great point - Turning down the 'amount' in the tone shaping block really helps. It's pretty harsh when the correction gets to be +20 DB in some spots. Taming that makes it sound much more reasonable and flat.

Probably going to make a video about this soon!

 

[kevinerror]

While the OP is super defensive, the reality, when all digital, the end the result, not the means, is all that matters. In each block there are EQ limitations and sound signatures, then the signal goes to the next block and is altered more, often in contradictory ways. Your 1k may have been up, down and sideways six different ways before the finally reading of it.

FWIW, a corrective (read: inverse) IR was all the range when the Verve 12MA FRFR was de jour in Axe world.

Can you expand on what you mean? Is there some altering of frequencies I'm not aware of when you run noise through the Axe..? I thought I was taking pure readings with all my blocks bypassed.

 

[Morphosis]

So measruement-signal -> filters -> TMA work better, than DEQ2496? In theory, it should ... smooth parameter can help not overcompensate ...

 

[kevinerror]

So measruement-signal -> filters -> TMA work better, than DEQ2496? In theory, it should ... smooth parameter can help not overcompensate ...

So far, yes. The DEQ worked just fine - again, I think 'close enough' with this is all one would need, little variances here and there aren't a huge deal - but the tone match block required less tweaking after the fact.

In the end, I am just using my ear to make those final tweaks, which I like to think is fairly objective and fair - but take that with a grain of salt, of course.

 

[Morphosis]

looking forward to your video with dtealid settings (especially of mic placement and filter settings & routing to tma) This was the reason i wished long time ago something like a tma block beside the global eqs ... cool, if this works for you ...

 

[RobinMatrix]

NL212 would be appreciated.

Yep, I'll do both the NL12 in 16R format, and the NL212 ... will need a fairly dry, windless day as I'll shoot the IR outside, so forgive me if it takes a few days, this is England, nice dry days are something of a rarity!

I can produce weak, normal and strong correction IR's with a bit of luck. Think of them not really as flattening, more sort of "de-flavouring" or neutralising the cab as much as possible ... a sort of audio rinse

 

[Morphosis]

Yep, I'll do both the NL12 in 16R format, and the NL212 ... will need a fairly dry, windless day as I'll shoot the IR outside, so forgive me if it takes a fee days, this is England, nice dry days are something of a rarity!

I can produce weak, normal and strong correction IR's with a bit of luck. Think of them not really as flattening, more sort of "de-flavouring" or neutralising the cab as much as possible ... a sort of audio rinse

Proud owner of a Q12a,not NL series - but this would be the best way to do it, IMHO! "de-flavouring" is, for sure, the right "title" - instead of "guitarcab-FRFR" or something ...