Tuesday, June 1, 2010

Impulse vs. Real Cabinet

I have always been a fan of modelers and hope that the next generation will be better than the last. It is a great theory, a whole room of amps and FX in a single software package or device. The problem is that they sound usable at best. Tube amps just dominate in so many ways.

However, the tube amps are large and heavy, require expensive maintenance, and are just a pain all the way around. Not to mention they are comparatively much more expensive even with offshore manufacturing closing the gap. Also in the studio (and even live) they are incredibly inconvenient. In order to sound good they must be loud and mic'd properly in a good room. Modelers can sound their best completely direct.

The reason I still support modeling is that I can record at anytime and anywhere with a laptop and USB interface. Also when mixing, a badly mic'd or improperly dialed amp can sound horrible. Or you need a specific tone that you don't have or don't have the time to mic, a modeler can be just the ticket.

Anyway I personally think one of the biggest breakthroughs and a weakness with modeling has been impulse responses of speaker cabinets. Essentially it is a sine sweep sampling. Do some searches and you will find that there are lots of links and discussion and packs for sale.

They sound very good, especially compared to other speaker simulation techniques, but they still lack something. Many say they don't breathe, that they are static and not 3D sounding. The problem is that most people don't really have the time to do a proper shoot out. Then using 3rd party impulses is a crap shoot since there is no way for you to compare with the real mic counterparts.

With free individual impulses it is often frustrating since there are so many out there of varying quality. Then with packs for sale or free, they have positions and combinations and such, but sometimes positions even down to a 1/2" is not enough. Millimeters in any direction or angle can have rather large effect on the tone.

But even for myself I have never done a proper shootout properly isolating all the variables... until now.

Testing Procedure

The goal here is to isolate everything but the speaker cabinet vs. impulse. So I mic'd up an ENGL 4x12 Pro (Celestion Vintage 30 speakers) with a single SM57. It is roughly at the edge of the cap where the cone starts, maybe 5mm more on the cone side. Straight on and about 1 inch from the grill.

The amp is an ENGL Powerball. The impulse was generated via the ENGL Powerball FX return. Master volume: 9:00, Depth Punch: 11:30, Presence 2:30. Sine Sweep and Deconvolved with Voxengo's Deconvolver 12-Seconds at 32-bit float 44.1khz.

Preamp is a TNC-73 Chinese Neve Clone from a group buy. They are similar to a Chameleon Labs. I had it set to low impedance and driving the it a little hard with the output rolled back.

DI's were cut with a Schecter A-7 7-string (mahogany body, maple neck, rosewood fretboard) , bridge pickup, Seymour Duncan JB7, tuned to standard with low B. DI was an EWI LDB-1 into a Behringer ADA8000 modded with BB OPA4134 clocked via a DBX Quantum mastering processor.

Re-amping was done via a DIY Jensen style Re-amping interface with an Edcor WSM 10k:10k.

The reamping was done with all the same settings through the mic'd cab. Then again but capturing the signal from the pre-amp and run through the impulse of the mic'd cab set-up loaded up in Poulin's LeCab, no high pass or low-pass filters. All variables isolated except the cab/poweramp combination and speaker impulse.


Each clip has 4-bars of 3 riffs double tracked on their own, then in a mix. Peak levels matched then adjusted slightly by ear.

16-bit 44.1Khz WAV Files. Below are 320kpbs MP3's.

ENGL Powerball Real Cab and SM57

ENGL Powerball Impulse

Interestingly I am having a hard time hearing a difference on my Roland MA-8's that are in my office. But in the studio the first thing that popped out was a increased high-end/upper mid-range and just sort of lacked depth and was just grindy.

From another perspective the Mic'd cab could be described as wooly and muddy and warm and distant. The Impulse being more clear and in your face. Plus people tend to like more high-end just like they like things louder and perceive distortion as being louder.

So to help minimize this perspective, I added a high-shelf to the Mic'd cab:

ENGL Powerball Real Cab and SM57 with a High Shelf

Oh how the plot thickens! Now the Mic'd cab to me has more buzz, but it has a crunch and a grind to it with a 3D-ness. The impulse sounds 2D and lacks depth, just something unpleasant.

So how do we add depth? Well besides EQing with the rest of the mix and using mids to move instruments, how about reverb? In this case the impulse should have grabbed the room as well. But maybe without the percussive nature of the sine sweep, we need a separate room impulse. So I threw on Cubase's REVerence with the default Studio Room Sample and blended to taste.

ENGL Powerball Impulse with REVerence

Future Tests

At this point it would be time for a blind test, and that will probably be for another time. My opinion is that you could work with this. To send this out blind without some "training" like this, it would be a total crap shoot, you can hear differences, but which is better? Which is the cab vs. the impulse. Maybe some new riffs and testing is in order.

Also I would like to take some modelers, plugins and hardware and do the same tests. Through a cab and then through an impulse of the same setup. Maybe with some forum support, this can happen. Although I will likely have to start over with a new impulse since I can't leave my cab mic'd up into eternity.

Other Arguments

One could argue that the tone is bad to begin with so it is impossible to make a fair judgement. Well that may be true, but that isn't the point, it is relative to each other. Also the playing is so bad... well yeah. I focused on what I do, lots of chunk, some thrash, and some diminished chords, I mean at some point every engineer should do their own tests and come up with their own conclusions.

With one mic (SM57) on the Powerball it is a careful balance. Usually I need to two mics, one getting mud and beef and a condenser for the highs. With a single 57 fizz is a problem, so I minimized that. But it is a bit muddy. Plus it needs some EQ as well. AKA. Mudball in some parts.

PHASE: Personally I could hear a big difference between the two on my monitors RAW, so I didn't try too hard to flip the phase. Once you get into processing, things get out of control. You can try on your own though since the guitars are panned hard left and right. But consider that there is a time delay between the speaker and the mic vs. the impulse. So you will have to compensate. Then you will want a frequency aware phase adjuster because of the time shift. Then since they are different takes, there will be some inherent clock skew which will give an exaggerated amount of left-over that will change over time. So go by the lowest amount.

At least for me this answers a BIG question: Are impulses a waste of time? I really wasn't sure. At this point I don't think they are, they are very close and definitely usable. Especially to the point where the average listener probably won't notice or care.

That being said I would probably still elect to mic a cab given the option.


teddy v said...

The impulse with REVerence definitely brings it close to the original mic'd cab to my ears (however I'm at work on cans and will need to check with monitors later).

Something about the overall warmth and spacial presence from the original cab that the impulse doesn't have, but still useful in my opinion for most home recording hobbyists doing demo work.

I've seen a couple of shootouts with plug-ins and modelers focusing only on high-gain comparisons, but would like to see a shootout more in-depth with cleans, slightly breaking up, overdrive, and high-gain... It would be a lot of work, but would showcase the modelers/plug-ins strengths and weaknesses (compared to real world gear) across many styles/users. How can I help with the process?

Great work, thanks for your time.

aortizjr said...

Yeah that would be great, but much harder. My biggest weakness would be access to lots of amps. I have a Fender Hot Rod Deluxe, a '70's Silverface Champ, but that is about all.

Also many vintage amps don't FX loops or master volumes for that matter, so running impulses might be tough.

And there is a lot more interaction and careful gain-staging between the guitar, player, and the amp. Versus metal which is paint shredding with sterile EMG's and the like.

Let me think about it and see what my time looks like.

Thanks for listening!

Pedrojoca said...

awesome test, but i'd really like the file that you used with REVevence! Is it an impulse?

aortizjr said...

I just used the default LA Studio one that comes up in REVerence. So I don't think I can even get the file out of it.