Category: Articles and Opinions

I was in the process of rebuilding my studio desk when it occurred to me that the slight change in height would affect the acoustics of the room.

“Well,” I thought. That means remeasuring everything. And an excuse to get a new measurement microphone.

I looked about and basically by accident stumbled across Sonarworks. They’re known for making headphone calibration software, and have recently begun a foray into monitor calibration. One of their packages was a demo, which came with a rather decent measurement mic, for a whopping $40.

What the heck -a cheap, decent measurement mic, and I can demo the software and see what it has to say about my room. Worst case, I end up with a measurement mic.

Okay, it ended up being a little less cheap than I expected, since the microphone shipped Fedex from freaking Latvia, so shipping charges were a bit much, but I was still ahead of the game. The mic arrived a few days later and I put it through its paces with Fuzzmeasure and REQWizard. Got some decent readings, noted a few trouble spots that I’ve always had near my crossover points and along a hard-to-trap z-axial node/null. I made a few adjustments to my subwoofer crossover point and levels, and got some improvement.

I’ve always been skeptical of room-correction/speaker-correction software. No matter how much fancy math one does, the cold hard translation to physical reality often gets in the way of idealized imrpovements.

But that didn’t stop me from firing up their “Reference” software demo.  Just to try it.

I am impressed.

Not without caveats, of course – I can imagine this working best in a treated environment and with rather good speakers, since it has to do a lot fewer crazy adjustments if we’re only talking about a few dBs here or there.  No speaker adjustment is going to adequately compensate for poor RT60 or flutter echo. Oh, and the documentation could be better.

The sampling process is pretty straightforward. You run the wizard, it chirps and sweeps from the speakers, and it makes adjuestments based on where the mic is at various points. It’s…distrubingly good at it, too. It guessed my monitor separation as being 3’11 inches. “That can’t be, I put my stands 4′ 6″ away from each other!” I got out my tape measure and…3′ 11″. Apparently all my turning and toeing and adjusting had cut 3″ from the distance on either side. So that was impressive. It also was able to quantify something I had noticed but couldn’t put my finger on – that one of my monitors was balanced slightly (1.5decibels) louder than the other. I went back and adjusted my monitor trim and a lot of stuff evened out sort of magically.

I re-measured, re-sampled, and then fired up the compensation plugin on a remix I’d been working on. The differences were subtle, but notable. My soundstage was deeper, my “sweet spot” was wider, and the detail on my recording was clearer. It’s easy to attribute this to some sort of crazy magic, but it seems like a pretty simple compensation – if my low-mid is too loud at certain points, that mud is going to obscure a lot. If I have a null at 140hz, I’m losing out on frequencies I should be hearing. Same with any phase adjustments affecting my stereo imaging – poor phasing can cancel out stereo information.

It is, however, fussy. I was moving some cables around and knocked a speaker out of place. I put it back without resampling/recalibrating, went back into my mix, and there were audible phase problems. I ran a re-calibration and everything went back to working fine. So it’s something I’ll have to be aware of. There are also strong level differences between calibrated and uncalibrated signals, even with the trim engaged and turned up the calibrated was still quieter than the raw signal. It’s not much of a problem, it just means I need to re-adjust my output levels for K-metering (and get better at using the plugin’s internal bypass instead of the DAW master bypass). Latency is also an issue in higher quality modes, and as is often the problem when using linear phase modes, there can be pre-ringing in low frequencies.

So, the big question is, though…does it help? I would assert that yes, it does. On the few mixes I’ve done with the demo, my first-pass output has been cleaner and more translate-able than those mixes I’ve done without it. Taking mixes to do the “car test” has been dramatic – usually I find myself going back and doing endless fiddling with the low end to try and get it to sit right in the mix, but both times everything came back clean on the first try. I’d love to ascribe the improvement to my mixing skills levelling up in some way, but I didn’t do anything different than usual, so it’s much more likely that being able to hear better was simply the ticket.

This is by no means comprehensive. I will continue to see how this works (I’m more than willing to pull the trigger on a purchase when my demo expires) and perhaps try multiple calibration tests to see if the results are similar each time. This really doesn’t speak to any of the other room correction applications on the market (IK ARC2, KRK ERGO, etc), merely a single instance of SonarWorks reference in a reasonably stable, treated environment. So mileage may vary, naturally. But dang, it seems to work much better than I expected.

About 5 years ago, Logic released “Space Designer” as a default plugin with Logic. While it wasn’t the first convolution reverb, it was the one that launched a reverb arms race – suddenly, convolution reverbs were all the rage, and everyone had to have one.  Not only did reverb plugins appear, but samplers and synths started coming with their own convolution reverbs built-in.

And you know what? Convolution reverbs are great. They’re awesome for creating realistic room sounds (and for doing other things, too. They’re remarkable flexible).

Of course, there’s one problem – realistic room sounds aren’t always good room sounds.  People still pined for the oft-grainy,  sizzly, but often lush sounds of an old-fashioned algorithmic reverb, because quite often, they just sounded nicely musical.  Certainly, hardware units by Bricasti and Lexicon weren’t exactly losing tons of sales.  Plus, finding an impulse-response recording for just the exact room you wanted wasn’t always easy.   And convolution can get processor intensive – not a problem if you’re recording a 5-track band, but for those with high track counts and differing spatial needs for each one, it could add up quick.

So the algo-verb started making a comeback.  Lexicon, unspurpsingly had a few entries in the market, as did Eventide.  Then players like Native Instruments got in on the game with their Softube partnership.  Many of these were emulations of classic hardware units. Most clocked in around $200.

Then a strange thing started happening, or at least, strange for the world of audio plugins.  A few vendors cropped up offering high-quality algorithmic reverbs, not modelled on any specific units, for startlingly low prices.

Two in particular have caught my eye – ValhallaDSP and AudioDamage.  AudioDamage has been around for a while, and are known for making a wide variety of unusual, interesting effects and synths that you didn’t know you needed until you bought them.  ValhallaDSP is just a guy who likes doing DSP.  The interesting crossover here is that AudioDamage’s algorithmic reverb plugin was in fact partially written by Sean Costello, who went off to found Valhalla.  So there’s some DNA shared here.

I did a quick shootout of several of these plugins, all grouped by the fact that they each cost $50.

ValhallaDSP ValhallaRoom: a more traditional room-simulation reverb.  It comes with a startlingly large variety of spatial algorithms, including some cheekily-named spaces called “Sulaco”, “Nostromo” and “LV-426.”  The user interface is large, colorful and friendly, and looks like something from a Star Trek control panel (that may not sound like a big deal, but a straightforward UI  makes a big difference in the time it takes to dial in a sound.

ValhallaDSP ValhallaVintageVerb: VVV is probably closer to the “classic gear” emulator.    In addition to its array of emulation algortihms, it has three separate “colors” – “now”, “1980s” and “1970s”, which simulate the reverbs of the respective eras by truncating the bandwidth of the reverberations and changing the modulation.  1970 is more band-limited than 1980 or now, and has a grainy quality; 1980 is darker with a slightly more digital edge to it, and “now” is smoother and fuller  but with less character.  This was clearly designed less as a “space simulator” and more as a “character” addition to a track.

AudioDamage Eos: the first of the low-cost algorithmic verbs to really it the virtual shelves, It combines a simple, straightforward UI with some novel approaches to reverberation.  It sports only three modes – P1 (mono plate), P2 (stereo plate) and “SH” (Super Hall).  I’ve used Eos a lot on the latest record,  and while superhall is the big selling point of the plugin, the Plate sounds remarkably good and has that lovely whooshy graininess that’s perfect for echoey trance stabs.

To my ears, ValhallaRoom comes out on top as the most useful “swiss army knife” reverb processor.  It’s clean, smooth and lush without being overwhelming or artificial sounding. VintageVerb is excellent, but inhabits a strange space of cool-sounding effects that won’t work for everyone, but work extraordinarily well when they do.  Eos is a ittle more bread-and-butter but also a bit harder to control and a slight bit more artificial-sounding; perhaps a better comparison would’ve been the superhall with the valhallaDSP “ValhallaShimmer” plugin, which is a more, well, shimmery Brian-Eno-ish large-space simulator.    Frankly, though, any one of these plugins are kind of ludicrously well-valued at $50.  They hold their own against  more expensive products and even sampled reverbs.  This does raise the question of just what you’re getting when you spend $200 for an algorithmic reverb plugin.  The cynic in me thinks the name on the badge is an extra $100 right there.

I recorded a brief snippet, a piano loop through each reverb with roughly the same settings – it’s impossible to match them 1:1 but I tried to come pretty close – large room, 10s decay, 30ms predelay, 50% mix, approximately the same cutoff and boost values.  First is ValhallaRoom, followed by VVV in each of “now”, “1980” and “1970” modes, followed by Eos, then Logic’s stock “PlatinumVerb” and Space Designer (using a 10s synthesized IR).

EDITED TO ADD: This has stunningly gotten picked up by a few places I didn’t expect to even notice it.  And, unfortunately has gotten a bit more, uh, hyperbolic language surrounding it than I expected either.  Just for the record, this is by no means a comprehensive shootout, nor am I comparing them directly with more expensive reverb units – just demonstrating that there are a bunch of good $50 reverbs out there that sound pretty damn good, and can hold their own against some of the more hyped ones. Nor is the audio file supposed to be comprehensive, just an example so you can hear the reverb tails.

[soundcloud url=”http://api.soundcloud.com/tracks/96027636″ params=”” width=” 100%” height=”166″ iframe=”true” /]