Introducing New 64-bit Amp Sims, EQ, and Room Modeller

[thecontrolcentre]

Thanks for sharing these

[donkey tugger]

Thanks a lot, Dave! Been admiring HeadCase for a long time now. BUT these ampsims bury my modest machine performance in an instant hopping to 110% (lol) CPU consumption in Reaper 5.40 x64 Win 10 x64 whereas other popular sims pale in comparison to it and here I mean the CPU. That's insane! I suspect a user error on my part though not sure what could be the cause. Your sims might be very demanding, too. Any thoughts?

Not getting 110% (!) but the amp sims are showing as quite high in FL; around 13%. As a comparison something like SoftAMP Flextron will use around 6-7%. This is on my old-ish i7 3rd gen laptop by the way.

Sounding nice though.

[DaveClark]

Hi Vasya,

The amp sims are somewhat demanding, but I'm getting results much more like dt just reported on an i7 3930K desktop machine from 2013 with Win 8.0 x64. Now if you modulate some of the parameters, you'll see vast increases in CPU usage because a lot of computation is being done as the parameters are changed. But normally that should not be the case.

The rotational room modeller is very CPU-intense; I plan to parallelize it but that will take awhile. I typically just work out those tracks and freeze them.

If you continue to see problems, please PM me and I'll send you an email address so we can get to the bottom of the problem. I may need you to send me a project file and so on.

Anyone else seeing high-CPU usage? Thanks for any info.

Regards,
Dave Clark

On Edit: I could dramatically lower CPU, but the sound would suck!

[Jafo]

Holy bleep, these are good! They have the response, clarity and warmth I've been looking for in an ampsim, without the sterile fizz or mud everything else seems to have. The flexibility of the tone stacks is a very welcome tweak. The usage is pretty high here -- 25% overall (100% on one of my four cores) on an admittedly pedestrian AMD FX-770K @ 3.5 GHz on Tracktion 5 with 256 samples (5.8 ms latency) -- but the results are fscking stunning. Add a switch for lower/realtime CPU and you'll make all my other ampsims sad and lonely. (I wouldn't mind a Class-A operation switch either, but that's just my naïve prejudices talking.)

[DaveClark]

Hi Jafo,

Thanks for the feedback!

CPU load reduction would not take much time, but testing would be excruciating to ensure that the quality didn't drop below my own requirements, so it won't happen any time soon.

Class A: Do you mean Class A power amp or something more general?

Regards,
Dave Clark

[DaveClark]

Hi all,

I've posted a tutorial video for the acoustic room modellers (RoomConvp and RoomConvpROT) on YouTube:

https://youtu.be/adlI0L8E7W8

This link is also at the Download Other page mentioned at the top of this thread in case I have to change/update this video; I may forget to update it here but should remember to update my web page.

Please remember that the ROT one is really CPU-intensive.

Regards,
Dave Clark

[zenvoxpop]

Hey, thanks for the reverb tutorial.

Yep, you are right the toggle switches work fine if you use them as buttons. My bad...
But, I've found another problem. In Reaper the presets are not recalled correctly when reopening the project
and the plugins always start at default preset.
Care to fix it?
I love the bundle otherwise.

[DaveClark]

Hi z,

Thanks for the problem report. I have verified that the project reload does not appear to reflect the parameter changes for RoomConvp but does reflect them for the other plugins for version 4.78 of Reaper (5.40 verified for RoomConvp only) for VST2. Interestingly I also found that some other hosts (using JBridge) also do not appear to reflect the parameter changes on reload. However, all my x64 hosts that are mainstream DAWs do work properly. Because of this, I'm not at all convinced that it's actually my problem. However, please do keep reading because there may be a surprise!


Stored parameters recalled for RoomConvp:

Samplitude Pro X3 Suite x64
SONAR X2 x64
FL Studio 12 x64
Waveform 8 x64
Mixbus32c v.3 x64
Mixbus4 x64
Renoise x64
Reaper 5.40 x64 for VST3 --- Surprise!

(It even worked for MM17 Premium x86 with 32-bit RoomConvp.)


Stored parameters MIA for RoomConvp:

PG Music BIAB + JBridge
PG Music RealBand +JBridge
Reaper 4.78 x64
Reaper 5.40 x64 for VST2
Podium 3.1 x64 + JBridge

Due to the "Surprise!" you can simply rename RoomConvp.dll (and/or any of the others, actually) to .vst3 and place them in the "EXE Consulting" folder of your favorite VST3 folder such as this:

C:\Program Files\Common Files\VST3\EXE_Consulting\RoomConvp.vst3

Hope that works for you as it did for me. You will, however, need 5.40.

Regards,
Dave Clark

[alteregoxxx]

Hi Dave,
just a quick test with all the amp sims and they seems to have a great potential: to my somewhat "trained ears", as guitarist and "electronic guy" for more than two decades, these are at least on par with the Mercuriall amp sims and, personally, I consider the latter the golden reference for circuit modeling accuracy: Tone and feeling on playing are really really good on your plugins!

[Rant mode on]
Mercuriall, to me, has the edge (by a large amount in a lot of cases...) on a lot of stuff made by big companies that fills their mouth with advertisment like "we 100% model this and that" despite having results that are often "unexiciting", to be polite.
Circuit design it's not my day-job but, as an electronic engineer and diy man, I often design "new" audio stuff for myself and this usually requires spending a lot of time performing also audio simulation of the design using LTSpice (Yes, LTSpice offers this really useful feature). Obviosuly, more than often, a comparison of my designs with "standard" tube preamps is needed, hence often I draw some of the "famous and hystorical circuits" in LTSpice and process some sounds trough them; let's immagine what I often hear when comparing these ACCURATE simulations with certain vst plugins that should accurately emulate that circuit...
[Rant mode off]

Now, on you ampsims, I confirm something is wrong with parameter storage in Reaper 5: renaming the plugins to vst3 solve the problem as you said.

Then, it seems something is wrong with the tonestack model "Classic_J" (JCM 800 to my ears, correct?):
it seems to have a strange "short slapback delay" effect on it, a sort of doubling effect. This effect is not present on the others tonstack models.

Tried also the room modeller: to my ears, it sounds really too thin. Comparing it to various Bricasti M7 impulses (to my ears the best reverb ever made), it lacks a lot of "bottom end" and depth.
Moreover it would be great triyng to model also the true stereo behaviour of a real space (like Bricasti M7 do): to me this contribute a lot to the realism of a reverb.

Last but not least, thank you for the plugs!!

[zenvoxpop]

Yup, renaming the files to .vst3 did solve he problem with presets.

These amps really suits my needs, guitar-wise, and are pretty close to what
I define as a perfectly realistic sounding amp simulator.

The thing I still miss is the compression of real amps. Recording of a real tube amp hardly need compressor
afterwards in signal chain, because of the natural compression (due to speaker characteristics,
tube configuration etc.). I still miss that in sims, It gives the guitar sound some sort of edginess
and dynamic definition.

Did you ever consider implementing some compression tools, maybe some opto stomp padal or something like that ?

[DaveClark]

Hi alteregoxxx,

Thanks for your feedback. Although you'll see that we have different tastes and opinions below, I really do appreciate your posting. It's good to see differing points of view.

Classic J is not JCM 800 but instead is James tonestack as I briefly (too briefly?) mentioned in the video.

If you analyze the room acoustic modeller, you should see that it is flat between about 25 to 30 Hz up to 20 kHz. It is not "missing a lot" in the low end unless you mean at 10 Hz or something like that. I don't want any EQ effects on my room modeller.

As far as "true stereo" goes, you may want to rethink that. Although you may like that sound, it's not very "true" but more like "double stereo."

On modelling real spaces: As I understand it, Bricasti M7 is a digital reverb that uses several independent models to simulate each component of a real space separately. When combined, the independent models have to be fine-tuned by hand to achieve anything like a real space. Unfortunately, it's just impossible to get all the exquisite timing correct using this approach. An acoustic problem is like a continuum that cannot be broken out into discrete pieces. Now you may like what they've done, but it's often not very realistic. I like this kind of thing only when I follow it with a simulation of a real space.

Based on your comments, it seems to me as though you prefer somewhat artificial sounding stuff more than I do. For cetain genres, it's perfectly OK, but for the stuff that I do, it's usually not OK.

Regards,
Dave Clark

[DaveClark]

Yup, renaming the files to .vst3 did solve he problem with presets.

<snip>

Did you ever consider implementing some compression tools, maybe some opto stomp padal or something like that ?

Hi z,

Thanks for getting back on the .vst3 rename.

I've thought about doing an opto stomp as well as other effects, but I've thought even more about NOT doing them! I already have a lot of this sort of stuff that seems to work well. Guitar amps and reverb I found not to be so good (for my music, anyway), so I worked on them instead.

Regards,
Dave Clark

[alteregoxxx]

Classic J is not JCM 800 but instead is James tonestack as I briefly (too briefly?) mentioned in the video.

See...I was doing other things simultaneously while watching the video, maybe I missed that part, sorry!
However, to my ears, the James tonestack gives the most JCM800ish sound of the bunch, more than using the tone stack Classic_M (that I suppose could be Marshall type; I know...I promise I'll see the video againg and concentrating only on doing that ).
Classic J however, as I already said, seems to suffer of a sort of "echo doubling" effect, a phenomena absent for the other tonestacks. Would you mind investigating on this?

If you analyze the room acoustic modeller, you should see that it is flat between about 25 to 30 Hz up to 20 kHz. It is not "missing a lot" in the low end unless you mean at 10 Hz or something like that. I don't want any EQ effects on my room modeller.
.....
On modelling real spaces: As I understand it, Bricasti M7 is a digital reverb that uses several independent models to simulate each component of a real space separately. When combined, the independent models have to be fine-tuned by hand to achieve anything like a real space. Unfortunately, it's just impossible to get all the exquisite timing correct using this approach. An acoustic problem is like a continuum that cannot be broken out into discrete pieces. Now you may like what they've done, but it's often not very realistic. I like this kind of thing only when I follow it with a simulation of a real space.

Based on your comments, it seems to me as though you prefer somewhat artificial sounding stuff more than I do. For cetain genres, it's perfectly OK, but for the stuff that I do, it's usually not OK.

Uhmm...I absolutely don't mean to be ungrateful, only trying to contribute; but I've never seen, in the bunch of real rooms that I measured over the years, including the treated ones of the recording studio that I have had for several years, one that was flat. I mean flat like the sea here in Naples in the perfect day of an italian August; and flat like the freq response of your room simulator.
Modal resonances and greater decay times at low frequency, typical of all the rooms that are not big as Trafalgar Square, combined with the greater absorption at medium and high frequency of the furniture that ALWAYS will be in room, even in minimal amount, will almost inevitably screw up the freq response, favouring low frequency build up. Even filling a room with hundreds of bass traps will not save you from this. Hence I guess we effectively have a somewhat different perception of what is realistic and what is artificial. I've not digged deep in the Bricasti M7 "behind the scenes" algos, and for sure you could be right on the independent models as being, "in principle", "unrealistic"; but I've also learned that is the final result that counts more than all. And I can assure you that certain Bricasti M7 presets, some rooms in particular, sounds really really really realistic, with the typical "full" ("bassy" if you prefer) and slightly attenuated hi-freq feel of all the good rooms I've recorded in. Something that is really missing in your room emu.

As far as "true stereo" goes, you may want to rethink that. Although you may like that sound, it's not very "true" but more like "double stereo."

Nope. I guess it's a matter of wording. With "true stereo" I mean this: try opening a send (post fader and post pan) from an audio track panpotted at center, towards a bus with your room sim on, then put in Solo this bus. You'll hear reverb coming from both side, left and right, as it should be. Now try panpotting the sending track full left and listen what come out from the bus with you room sim on: nothing comes out from the right channel. This is not realistic because, in a real room, even if you place an instrument on the left wall and play it, there will always be reflections coming also from the right wall.
Moreover, for sure you already perfectly know this stuff, because in your room sim plugin there is a parameter just to choose the position of the source in the room, and it behaves exactly as I've described.
I was just saying that is impratical in a mix section to use the plugin that way, it's a lot more pratical if the reverb automatically follow the panning position of the source, without having to change any reverb parameter. That's all. And that is the way BricastiM7 (but is just an example, all true stereo units behaves in this way) reacts.

[DaveClark]

Hi alteregoxxx,

Thanks for your response. I'll attempt to answer some of the point you've raised now that I see better what you intended.

Classic J however, as I already said, seems to suffer of a sort of "echo doubling" effect, a phenomena absent for the other tonestacks. Would you mind investigating on this?

I appreciate your having mentioned this. When I get back to looking at tonestacks, I'll try to remember this and take a look --- may be some time. I have done a few tracks with the Classic J and didn't hear anything wrong with it, mainly just making sure that the tone controls worked and that the sound was good overall.

Update (11 August 2017): I did indeed finally hear the problem described when the treble control is turned down with the James (Classic_J) tonestack. This is fixed in version 0.0.3.0. There actually was a problem with all tonestacks, but it was audible only with the James. I plan to upload the new versions to KVR Audio, but for now:

http://www.execonsultingtx.com/download/download-other/

Modal resonances and greater decay times at low frequency, typical of all the rooms that are not big as Trafalgar Square, combined with the greater absorption at medium and high frequency of the furniture that ALWAYS will be in room, even in minimal amount, will almost inevitably screw up the freq response, favouring low frequency build up.

Quite right! I specifically mentioned in the video that I did NOT model the accelerated decay of high frequencies. The reason for this, as I said in this thread, is that I want the room response to be flat as can be. I'm not trying to model the back seats of a large concert hall on a humid day or anything remotely like that, and I don't want to have to re-EQ each track in a mix just because I suddenly turn on room modelling. I'm modelling what I regard as ideal positions in an idealized room that does not exhibit accelerated high-frequency decay.

Now I have to admit that many years ago, an earlier version of this program did model the effect you are talking about, but I didn't like it at all, so I took it out.

... slightly attenuated hi-freq feel of all the good rooms I've recorded in. Something that is really missing in your room emu.

Yes, purposefully missing as I just wrote above.

Something that is missing from all other emulations except mine, as far as I know, is the modal decay. In a real room, frequencies are decaying in an accelerated manner not just at the high end, but all throughout the spectrum. The spectrum gets thinner as time goes on and some modes decay into others. Did you ever consider this effect? Do you know what I'm talking about? This is the effect I prefer to model.

Now try panpotting the sending track full left and listen what come out from the bus with you room sim on: nothing comes out from the right channel. This is not realistic because, in a real room, even if you place an instrument on the left wall and play it, there will always be reflections coming also from the right wall.

Panpotting is what is not realistic --- seriously. You cannot move a musician in a room by turning a knob. Many things change as you move an instrument around in a room, not just the relative volumes. For example, moving includes changes in coloration, some very pronounced. What you are doing with panpotting with my room modeller is something that should absolutely NOT be done with my room modeller as I tried to say in the video. Panpotting is "Evil."

A practical consideration of implementation: Allowing continuous panpotting of a source would require (to be accurate) a continuous set of solutions to the wave equation across the entire semicircular arc around the listener. That doesn't seem reasonable to me (and I don't want to interpolate).

Again, thanks for your comments and I hope that I've addressed them in an appropriate fashion.

Regards,
Dave Clark

[PeterPan8]

The Download page opens an ftp page where you could (supposedly) download the files. But right-clicking on the file to download takes on forever, and does nothing.

I closed the page and tried again from the main page, but your website page doesn't open.

Maybe a fix is due!