The “problem” with impulse response for amp cabinet sounds

[Marco Raaphorst]

Using an impulse response might not be the best method for simulating the sound of a speaker cabinet in my opinion. You can read it on my blog:
http://melodiefabriek.com/blog/problem- ... -response/

[DJDark2010]

Hey Marco,

I just purchased your Rockman refill.
Also great blog article. Thanx.

[Marco Raaphorst]

Hey Marco,

I just purchased your Rockman refill.
Also great blog article. Thanx.

Thanks for buying Rockmen!

I am planning to write some more articles about Reason and saturation. I love saturation

[normen]

In your article you kind of bounce back and forth about speaker breakup. First (and finally) you say that the breakup adds nice tone but when you talk about the Palmer box you say that the lack of breakup adds dynamic which you find nice as well.

Additionally, the "sound of a room" does NOT change with volume. One might think that a higher volume excites the room in a different way when listening but only because the reverb gets louder as well (completely linearly with the source signal volume). In this respect an impulse response is a perfect replication of a certain speaker/mic setup, at any volume. The only factor here is speaker (and mic) breakup, that is distortion from overloading the coil mechanically.

Furthermore "cab compression" is basically only that, speaker breakup. Its not "compression" but pretty much distortion - which obviously has a compression characteristic. But people tend to under-estimate the compression built into the ears when turning up an amp like mad and trying to judge whats happening sonically. Any additional compression is mostly amp sag.

Not trying to bash you here or anything, just giving some feedback/info - I like your blog posts and want them to keep high quality

[Marco Raaphorst]

In your article you kind of bounce back and forth about speaker breakup. First (and finally) you say that the breakup adds nice tone but when you talk about the Palmer box you say that the lack of breakup adds dynamic which you find nice as well.

Yes I like that tone as well. It's not the same as an impulse response though. The sound of the Palmer is very open, something you can only achieve using filters/eq imo.

Additionally, the "sound of a room" does NOT change with volume. One might think that a higher volume excites the room in a different way when listening but only because the reverb gets louder as well (completely linearly with the source signal volume). In this respect an impulse response is a perfect replication of a certain speaker/mic setup, at any volume. The only factor here is speaker (and mic) breakup, that is distortion from overloading the coil mechanically.

Have you never noticed how volume causes a room to resonant differently? Below a certain point, no issues, above it things start to change. Some material works as an acoustic compressor, absorbing sound at certain levels and frequencies.

Furthermore "cab compression" is basically only that, speaker breakup. Its not "compression" but pretty much distortion - which obviously has a compression characteristic. But people tend to under-estimate the compression built into the ears when turning up an amp like mad and trying to judge whats happening sonically. Any additional compression is mostly amp sag.

Not trying to bash you here or anything, just giving some feedback/info - I like your blog posts and want them to keep high quality

Their is compression caused by a speaker cabinet. Compare a closed back to an open back speaker cabinet. It has everything to do with air pressure inside the cabinet that gets compressed and effects the speaker. A difference in sound simply caused by the way air flows. Even drilling one hole in a speaker cabinet creates a difference in sound pressure. That's not break-up, that is air compression imo. Or maybe should called cap compression

[normen]

Have you never noticed how volume causes a room to resonant differently? Below a certain point, no issues, above it things start to change. Some material works as an acoustic compressor, absorbing sound at certain levels and frequencies.

[...]

Their is compression caused by a speaker cabinet. Compare a closed back to an open back speaker cabinet. It has everything to do with air pressure inside the cabinet that gets compressed and effects the speaker. A difference in sound simply caused by the way air flows. Even drilling one hole in a speaker cabinet creates a difference in sound pressure. That's not break-up, that is air compression imo. Or maybe should called cap compression

Both of these effects are completely linear in terms of volume. They are less noticeable at lower volumes, but it doesn't change their characteristic.

[Marco Raaphorst]

Have you never noticed how volume causes a room to resonant differently? Below a certain point, no issues, above it things start to change. Some material works as an acoustic compressor, absorbing sound at certain levels and frequencies.

[...]

Their is compression caused by a speaker cabinet. Compare a closed back to an open back speaker cabinet. It has everything to do with air pressure inside the cabinet that gets compressed and effects the speaker. A difference in sound simply caused by the way air flows. Even drilling one hole in a speaker cabinet creates a difference in sound pressure. That's not break-up, that is air compression imo. Or maybe should called cap compression

Both of these effects are completely linear in terms of volume. They are less noticeable at lower volumes, but it doesn't change their characteristic.

Even super hifi speakers are imo never linear. The sound characteristics are changing depending on the volume level. But if I understand you correctly you think this is not true?

[selig]

Their is compression caused by a speaker cabinet. Compare a closed back to an open back speaker cabinet. It has everything to do with air pressure inside the cabinet that gets compressed and effects the speaker. A difference in sound simply caused by the way air flows. Even drilling one hole in a speaker cabinet creates a difference in sound pressure. That's not break-up, that is air compression imo. Or maybe should called cap compression

Sources? I can't find any data to back up this assertion…First problem to what you've said above is that "air flows" (wind, a fan, etc.) which is true, BUT that's not how SOUND works. Sound doesn't move air, it modulates air pressure. So air can flow but sound does not.

As for your assertion that drilling a hole in a sealed cabinet changes the sound, this may be true - but your CONCLUSION that it's caused by "compression" is not backed up by science. It's also true that a sealed room sounds/measures differently when you open a door, but that doesn't mean the difference is because the room is "compressing" in any way. I would suggest it's more about resonance and standing waves rather than compression of any sort.

As for closed back vs open back, again a closed room sounds different than a room with one wall removed - it's changing the resonances/standing wave response, not the "compression". Louder sounds do not produce more air pressure, they only modulate that air pressure across a wider range of air pressures both positively and negatively. The only way to change the overall air pressure in a sealed room would be to ADD more pressure from the OUTSIDE, unless I'm mistaken. But if the room/cabinet is SEALED, this is impossible.

Other possible factors…
There is a common data point quoted online that says a sound cannot be louder than 194 dBSPL, which is pretty damn loud. But the additional facts are this: a SINE wave cannot be louder than 194 dBSPL WITHOUT DISTORTING. So yes, sines CAN be louder than 194 dBSPL, but may distort. But here's the kicker - the REASON a sine cannot be louder than 194 dBSPL without distorting isn't because of some limit to the waveform PEAKS, it is because the waveform TROUGH cannot be any LOWER because it is at a VACUUM at that level! This physical limit is at the BOTTOM of the waveform, not the top.

What is also important here is that at these extreme levels, it is not the space that is "compressing" the sound and keeping it from getting any louder, it is the inability of air pressure to go "negative" so to speak. This same 194 dBSPL limit will exist equally in a closed space as it exists in open air, at least as far as my research has shown. I can find no data to support the idea that a small enclosed space will "compress" the sound at some level. If this IS true, than there will be a relationship of the volume of the space to the compression level, the "threshold" if you will - and I can find no discussion of any of these concepts nor have I ever heard of them in the past.

Again, if you have sources that say otherwise, I'm happy to stand corrected. All I'm saying is I've searched for data to back up your claim that an enclosed space will compress the sound at a certain level, and I've found none.

I must say I find this discussing fascinating, and would love to learn something new here about how sound behaves - it's an important subject for audio engineers to be sure!













Sent from my iPad using Tapatalk

[selig]

Have you never noticed how volume causes a room to resonant differently? Below a certain point, no issues, above it things start to change. Some material works as an acoustic compressor, absorbing sound at certain levels and frequencies.

[...]

Their is compression caused by a speaker cabinet. Compare a closed back to an open back speaker cabinet. It has everything to do with air pressure inside the cabinet that gets compressed and effects the speaker. A difference in sound simply caused by the way air flows. Even drilling one hole in a speaker cabinet creates a difference in sound pressure. That's not break-up, that is air compression imo. Or maybe should called cap compression

Both of these effects are completely linear in terms of volume. They are less noticeable at lower volumes, but it doesn't change their characteristic.

Even super hifi speakers are imo never linear. The sound characteristics are changing depending on the volume level. But if I understand you correctly you think this is not true?

I read Normen's comments to refer to room resonances and cab compression. Both of THOSE effects ARE linear in terms of volume. He was not talking about DRIVERS/SPEAKERS - you were talking about CABINETS compressing, not drivers breaking up, right?






Sent from my iPad using Tapatalk

[jengstrom]

Actual distortion at loud volumes, where the speaker cones start flapping about and making their own mechanical side-noises and such types of distortion tones, can't be repeated with any precision by using impulse response recordings. E.g. the tone of a "nicely" broken speaker.

Impulse response is perfect for replicating any (frequency-specific) delays and well as EQ and volume level attenuation.

@selig: Really interesting about the pressure modulation limit at air pressure zero - I had never thought of that. That also means if the ambient air pressure is changed, that limit moves - sounds can be louder (in dBSPL) at higher ambient pressures.

[normen]

Selig explained it perfectly. Sure a speaker/amp combination has "compression" characteristics but they are defined by the electric limits of the amp and the physical limits of the swinging coil. You hear this myth about rooms sounding different at different volumes every once in a while but I pointed out where that impression might come from. It might even sound reasonable to common sense but the fact remains that the plural of "anecdote" isn't "data" - and theres no data to back this up. There is a lot of data on the fact that rooms (or small wooden rooms a.k.a. cabinets) have the same characteristic at any level though.

[Marco Raaphorst]

Interesting!

Selig, is in your opinion an impulse response capable of totally capturing the sound of a speaker? If everything is totally linear it should be capable to do so.

In my experience all amp simulators have "certain dynamics" which are sounding different than a speaker which is mixed. And the weird thing is, no one sounds like my Palmer also. But maybe what I am hearing is still harmed by the fact I think I know what I am hearing?

I do believe a speaker, any speaker (I have listened to some of the best, electro static ones), is not truly linear. But maybe I am mistaken.

Yes, interesting. In the end I am not a theoretical person. All I am doing is based on what I would like to hear.

[Marco Raaphorst]

Selig explained it perfectly. Sure a speaker/amp combination has "compression" characteristics but they are defined by the electric limits of the amp and the physical limits of the swinging coil. You hear this myth about rooms sounding different at different volumes every once in a while but I pointed out where that impression might come from. It might even sound reasonable to common sense but the fact remains that the plural of "anecdote" isn't "data" - and theres no data to back this up. There is a lot of data on the fact that rooms (or small wooden rooms a.k.a. cabinets) have the same characteristic at any level though.

Thanks. And can "we" go as far as saying an impulse is great for capturing the sound of the room?

Thanks alsof for clarifying the "electric limits of the amp". This makes sense to me at least, I think/believe, this is what I am hearing, this randomness which makes a guitar amp sounding so musical.

[normen]

Thanks. And can "we" go as far as saying an impulse is great for capturing the sound of the room?

As I said before, it perfectly captures a static combination of room, mic, mic position and sound source position. Obviously it doesn't capture people moving in the room, the sound source moving or even artificial reverbs which often modulate some parameters very well.

[selig]

Interesting!

Selig, is in your opinion an impulse response capable of totally capturing the sound of a speaker? If everything is totally linear it should be capable to do so.

In my experience all amp simulators have "certain dynamics" which are sounding different than a speaker which is mixed. And the weird thing is, no one sounds like my Palmer also. But maybe what I am hearing is still harmed by the fact I think I know what I am hearing?

I do believe a speaker, any speaker (I have listened to some of the best, electro static ones), is not truly linear. But maybe I am mistaken.

Yes, interesting. In the end I am not a theoretical person. All I am doing is based on what I would like to hear.

An impulse response should be perfectly capable of capturing the linear aspects of a cabinet, but obviously not speaker breakup etc.
BUT, amp simulators are not the same as cabinet simulators because amps are not linear.

IMO you are mixing up "speaker" with "cabinet". A speaker is NOT linear, but the cabinet IS linear. Therefore a cabinet cannot compress, but a speaker can breakup.

In the Ideal emulation world we should separate the speaker from the cabinet and model each separately. In that scenario the cabinet would be linear but the speaker would not. This would solve the issues we're dealing with. This also implies that an IR would not be the ideal way to model a cab/speaker because you can't take an impulse response of a cabinet that has no speaker!

The next generation of modeling will have to move beyond IRs to physical modeling IMO. But IRs will still be useful for gathering certain data or for simpler "models" that require less CPU.






Sent from my iPad using Tapatalk

[Marco Raaphorst]

Interesting!

Selig, is in your opinion an impulse response capable of totally capturing the sound of a speaker? If everything is totally linear it should be capable to do so.

In my experience all amp simulators have "certain dynamics" which are sounding different than a speaker which is mixed. And the weird thing is, no one sounds like my Palmer also. But maybe what I am hearing is still harmed by the fact I think I know what I am hearing?

I do believe a speaker, any speaker (I have listened to some of the best, electro static ones), is not truly linear. But maybe I am mistaken.

Yes, interesting. In the end I am not a theoretical person. All I am doing is based on what I would like to hear.

An impulse response should be perfectly capable of capturing the linear aspects of a cabinet, but obviously not speaker breakup etc.
BUT, amp simulators are not the same as cabinet simulators because amps are not linear.

IMO you are mixing up "speaker" with "cabinet". A speaker is NOT linear, but the cabinet IS linear. Therefore a cabinet cannot compress, but a speaker can breakup.

In the Ideal emulation world we should separate the speaker from the cabinet and model each separately. In that scenario the cabinet would be linear but the speaker would not. This would solve the issues we're dealing with. This also implies that an IR would not be the ideal way to model a cab/speaker because you can't take an impulse response of a cabinet that has no speaker!

The next generation of modeling will have to move beyond IRs to physical modeling IMO. But IRs will still be useful for gathering certain data or for simpler "models" that require less CPU.



Sent from my iPad using Tapatalk

Yes I am mixing up "speaker" with "cabinet". I mean the sound of the speaker. I took the blogpost offline. Needs a complete rewrite

It is indeed the non linear speaker I am missing in all speaker sims I have heard. They have a tone and feel that is different. And that has a lot to do with dynamics imo.

[normen]

Many of the modern sims actually do emulate speaker breakup. E.g. Eleven Rack and AxeFX even have a separate parameter for that so you can "age" or "break" your speaker with a slider. With modern amps most of the distortion happens in the preamp stages however, speaker breakup gets less and less important the further you move away from the old Fenders and Voxes of this world. The modern Soldano/Mesa/Engl sounds have nothing to do with speaker breakup really, they operate the speakers mostly in spec and sound the same at any output volume - except for guitar feedback etc. caused by high volume of course. Actually the distortion sound is so intricately designed these days that any additional "random" distortion would be undesirable.

But yeah, back when amps weren't really designed to distort people got distortion any way they could, including overloading their speakers.

[Marco Raaphorst]

I love that "old" sound. That unpredictable sound. It's warmer imo. That Stevie Ray Vaugh tone. Strat in Neck or Middle pickup and getting that fat but bell like tone. I tried so hard to get this tone using any amp sim available. This is why I decided to roll my own patches and see if I could correct some of the issues which are in Softube Amp etc.

[cosignsessions]

I'm just gonna read and eat my popcorn. This topic is of great interest to me.

[Marco Raaphorst]

the post should be fine now http://melodiefabriek.com/blog/problem- ... -response/

thanks all!

[per-anders]

The "problem" with IR's for amp cabs is that it doesn't sound or feel right. Don't need to go any further than that really unless you want to try and analyze why for the sake of making something better.

Even if it did nail the sound as recorded it still wouldn't sound right to most guitarists, because they're expecting "amp in the room".

I'd love for there to be a better option, but right now there isn't. So while you may not be able to completely nail that classic tone ITB if that's your only option for recording then all you can do is try to make something that sounds good to your own ears using what's available, and to be fair I don't see any of these modern tools as a handicap in that endeavor.

[mooseharris]

Interesting thread and an interesting blog post.

I'm just getting into the whole impulse response thing, and have spent the day working on a sim of my live set-up. I've used a 1974 Hiwatt DR103/1988 Ampeg SVT810E combo for bass since the late '80s (used to have a double stack, but these days it's excessive for the venues I play). I've used IR to emulate the Ampeg, whilst approximating the Hiwatt using a tweaked version of the Vintage P-Bass patch in Cerberus.

The Cerberus 8x10 sim isn't as accurate as using the IR files I got from Red Wire. I opted for six mic positions: cone/cap at 0", room at 12" and distant, ribbon mic angled at 45o 12"/24" above the cabinet. I then went for two different mic sets: Shure SM57, Earthworks TC30 and Coles 4038 or ElectroVoice RE20, Neumann U47FET and Royer R121. The combi I put together allows you to choose mic position and volume for each set, and has a DI signal which can be added to the mix. I've arranged the Hiwatt patch so the Ampeg sim can be added, and switched between either the built in Cerberus cab simulation or the additional Ampeg combi. Takes a bit of patching knowledge, but seems to function fine. There are also some backdrops for each bit of gear.

I've found it gives a decent emulation of my live set-up, but as Marco suggests, it's not the same as the real thing.

When I used to record professionally in the '80s and '90s, digital gear was in its infancy. We had the AMS RX16 and the early versions of the Lexicon, like the PCM 42 and the Prime Time, but not much else. I was lucky enough to work with a couple of producers who learned their craft in the '60s and '70s, Glyn Johns and Tom Dowd. Whilst both readily embraced technology, they still had their old methods to fall back on. This is where some of what Marco was saying comes into its own.

To do a tight ADT-style delay, we used to stick the Ampeg or a Marshall 4x12 in the studio toilet. The tight echo made for an excellent double track effect. I've done an IR of my own toilet (bare plaster walls, tiled floor, so plenty of reflective surfaces) which sounds pretty good. The problem is the IR doesn't react to increased volume the way the room does.

Cranked up, the 8x10 cones rattle. A couple are slightly damaged, and this adds to the sound at volume, which is why, in 25 years, I've never replaced them. The valves in the Hiwatt rattle happily when it's stacked on top of the cabinet, and the cistern in the toilet, the wash basin and the metal bin all rattle too. IR doesn't give you all this.

If I'm roughing out songs I'll use a sim and IR, and have even added sampled valve hum in the quiet bits, but if I'm recording properly I re-record or re-amp the DI signal of the bass through the Hiwatt in the toilet, because it's more natural, and more real.

There are pros and cons to both, especially that of not humping a massive bass rig around the house and running yards of cable between rooms, but overall, whilst I'm enjoying experimenting with IR, I think the old methods have a more pleasing final result.

If I can work out how, and if the Red Wire license permits, I'll add the IR and combinator files.

Here are the backdrops:

Hiwatt.jpg

Ampeg.jpg

[Marco Raaphorst]

Interesting. My Marshall rattles also. Something I don't like.

I am wondering how Kempler works. They create IRs it seems in a glever way. Not sure how it works and how succesful it is in capuring the tone. Some things I heard sounded great.

[mooseharris]

I guess rattle and hum are a thing of preference. I have a low action on my '59 Precision expressly to achieve fret rattle. Coupled with cabinet and amp rattle it tanks along nicely at high volume, so is great for gigging. My '95 Precision is set up to be less rattly, and gives a more precise sound when recording. I set my guitars up in a similar way.

[jonheal]

To sidetrack a bit, I remember from high school chemistry (a LONG time ago) the teacher telling us, "gases are so energetic, even one molecule effectively fills the room." I've thought about that from time to time, and today I thought I would actually look it up. Turns out, the average "room temperature and pressure" molecule of "air" is crashing around the room at about 1,000 MPH!

http://scienceline.ucsb.edu/getkey.php?key=193

So even though I tend to think of air as a sort of stable medium though which the wave energy of sound propagates, in reality, the medium itself is churning on its own in an unfathomably complex fashion.

I wonder how many 1,000 MPH air molecules have to slam into your eardrum before you hear anything?