Spitfire: Air Studios Reverb

[Guy Rowland]

AIR Studios reverb is our flagship effects plugin. A convolution reverb with 67,000 impulse responses, True Stereo signal paths and Virtual Positioning Technology.

Hall Controls
Modify physical attributes within the hall to control the reverb tone and decay. Settings include:
Hall Layout: Use this control to invert the room configuration. This affects both microphones and player position.
Canopy Height: Here you can set the canopy to 1 of 3 different heights: Low, Medium or High. Lower values have a smaller reverberation time than higher values
Canopy Material: Choose from 2 different materials on the canopy: Wood or Melatech. Melatech will give a softer, less reverberant sound than Wood.
Gallery Dampening: Enable this to add absorption into the gallery portion of the hall. Adding this will reduce the reverb time.

BOTTOM BAR
Input: Set the intensity of the input signal and the size of the ensemble, with gain control, mono/stereo switch & width of the source/ensemble.
Reverb: Augment the sound with artificial pre-delay, and quick access to early and trail stretch controls.
Output: Final tweaks to the balance between Dry & Wet signals and the output gain

Virtual Positioning Technology
Drag and drop your sounds and musicians anywhere in the legendary Lyndhurst Hall. Choose your sound to be either mono or stereo.

Mixer
Adjust your microphone signals to a variety of settings including tree, outriggers, binaural, ambient, gallery wide and gallery far.

Source
Modify the performers’ radiation in the room, and which way the source is facing. The controls operate all within the rotation control.

Shape
Augment the reverb at various points in its overall decay time. Examples may be turning the Hall into that of a chamber, or alternatively creating huge wash reverb for synths.

EQ
Use the 5 band graphical EQ to either shape the input signal going into the Hall, or the Hall reverb itself by placing it pre or post.

Preset Browser
Search for specific instruments as well as some presets created by Spitfire Audio and industry leading engineers

£279 intro until June 16, then £349

Surprised there's no surround for this, but I quite like what I hear. An A/B with the Kontakt sampled mic positions would have been interesting.

[Luciano Storti]

Oh My. They actually did it.

I reckon the advent of Berlin Studio helped nudge everyone involved into this direction. Cool to have the option.

Blasphemy alert: I don't revere the sound of Air Lyndhurst as much as many others.

[Jaap]

I join you in that Luke. Though I have nothing against it, I never really connected with it as well and the whole fandom about always fascinated me.

Price is very high in my opinion, specially without surround or a demo version to check cpu usage.

[Linos]

Knew that it was coming and have receoved an NFR. I think it is very well done and complete. After all ot's the first plugin that they have developed.

The shape parameters make a lot of sense and it's not something I have seen anywhere else. Adjusting the radiation pattern depending on the instrument does a lot for the sound.

I agree that a surround option would be nice. Paul wrote that they are looking into it.
I also get the reluctance regarding Airs sound signature. It is a lot of ambience that you add with this reverb. Even just the tree mic has a lot of depth and ambience. I am experimenting with how to make this reverb a bit subtler. I'm not there yet.

[Luciano Storti]

I agree that a surround option would be nice. Paul wrote that they are looking into it.
I also get the reluctance regarding Airs sound signature. It is a lot of ambience that you add with this reverb. Even just the tree mic has a lot of depth and ambience. I am experimenting with how to make this reverb a bit subtler. I'm not there yet.

If you are so inclined, it would be great to hear the fruits of your labor when you're ready - I seem to remember you like your mockups/recordings fairly dry?

I took the time to watch the walkthrough so I could reasonably comment, and it does look quite flexible and with lots of control, for Impulse Responses.

It makes me think that we might see more of this trend in the future: specific halls as one plugin, I mean. Studios have been done as single-function plugins for a while (UA, IKM) but not larger halls I think (I guess Teldex/Berlin Studio qualifies). Anyway, it's just really surprising to me that Air Studios went for this.

[tack]

Price is very high in my opinion, specially without surround or a demo version to check cpu usage.

Agreed. Reverbs are commonly priced far beyond what I consider sane. On the other hand, the gimmick itself (impulse responses captured at Air Lyndhurst) is not something that's easy to come by, so from a scarcity perspective I can see paying a premium. (Not that I'm personally willing to pay that premium.)

The xylophone example in Paul's walkthrough was interesting. I pulled up the xylophone patch from Spitfire Percussion, which was of course recorded in the hall, and it was extremely close.

[Guy Rowland]

Price is very high in my opinion, specially without surround or a demo version to check cpu usage.

Agreed. Reverbs are commonly priced far beyond what I consider sane. On the other hand, the gimmick itself (impulse responses captured at Air Lyndhurst) is not something that's easy to come by, so from a scarcity perspective I can see paying a premium. (Not that I'm personally willing to pay that premium.)

The xylophone example in Paul's walkthrough was interesting. I pulled up the xylophone patch from Spitfire Percussion, which was of course recorded in the hall, and it was extremely close.

That's good to know, thanks Jason.

For the longest time now, the received wisdom has been that samples must be recorded in great rooms because you can't emulate them properly.. It's always mystified me - surely we should be able to emulate it by now? I hate the baggage that comes with large sample libraries, the trillions of gb to install them and the imperfect legatos. I'm still attracted to the model of basic close mic set up, using fx for the room.

So I'm going to be following this closely. CPU and real world experiences are particularly interesting to me.

[Piet De Ridder]

(...) I'm still attracted to the model of basic close mic set up, using fx for the room.

The sound of acoustic instruments, certainly orchestral ones, is for a significant part generated by the instruments’ interaction with the room. I mean: an important percentage of the ‘complete’ sound comes from the room, not directly from the source. Adding reverb to a close recording of a brass section almost never gives good results because you’re adding reverb to “only half the sound”, the other half (which the room generates) isn’t there. Sound blooms in a room, if that bloom isn’t there, artificial reverb, on its own, can’t create it because it doesn’t have anything to work with. Not even SPAT, which shrinks from very few spatial challenges, can do it.

(I imagine that, within a few years time, AI-based spatializers will start providing acceptable solutions: when they know what they’re processing and they’re fed all the required information regarding the illusion you expect them to mount, they’ll be able to generate the correct bloom accordingly.)

This bloom problem is not so big with woodwinds (though it’s there too) because these sounds usually have not enough energy and impact to begin with (and thus don’t trigger the room as much), but with the three other orchestral sections — brass and percussion especially — you really do need more than a ‘basic close mic’ setup for a good, rich, convincing orchestral sound.

There’s no way you can create Andy Blaney’s Spitfire/Lyndhurst sound with, for example, pre-Synchron VSL libraries. It’s impossible. (You can’t do it with Synchron libraries either, but that’s got nothing to do with the bloom problem.)

The same thing, be it on a smaller scale, is also very noticeable with (good) drum samples. Compare the close mic signal of a snare to that in the overheads or the room mic’s: huuuuuuuuuuuge difference. (And the same goes for all the other drum instruments: kick, toms, hi-hats & cymbals.) The close mic of a snare has the crispness and the detail but it usually also sounds small and a bit weak, while the whacking impact and full size of the snare’s sound has to come from the more distant mics. You really need both for a complete snare sound. And again, I know of no existing reverb/spatializer than can generate a fully believable overhead/room signal from just a close mic’ed signal.

__

[Guy Rowland]

All true Piet, but where I remain eternally confused is that this phenomenon does not work by magic. The room turns that thin sound into the big bloom by physics. Can that process truly not be modelled?

I appreciate that modelling is not how Air is designed, it's an awful lot of impulse responses with clever ways to control them. The end result sounds pretty darn good to me, all the limitations notwithstanding.

[Piet De Ridder]

All true Piet, but where I remain eternally confused is that this phenomenon does not work by magic. The room turns that thin sound into the big bloom by physics. Can that process truly not be modelled?

Not with any of the software I’m familiar with, no. A definitely not with today's convolution-based spatializers because they rely for their tasks almost entirely on the information captured in their own samples, and hence are limited to what that information covers.

UA, IK Multimedia and Softube all have spatializers/reverbs for specific rooms: UA’s Ocean Way and Sound Studios, IK’s Fame and Sunset Studios and Softube's Atlantis Chambers (my favourite of this lot) and all of these are largely (or entirely, I’m not sure) modelling based, thus: algorithmic devices.
But again: I'm not convinced by the results of processing dry acoustic sources with these plugins and the quality of bloom they generate. They're all great for more traditional reverb usage though.

Anyway, the algorithmic approach makes the most sense to me because another reason why I have doubts that IR’s are the future, is because they’re static things. I haven’t watched the entire walkthrough of the new Spitfire plugin but I wonder: is it possible, for example, to automate changes in location and direction of projection without the software stuttering and crackling? In SPAT that’s perfectly possible, in real time, and the software doesn’t even blink. Convolution-based spatializers, on the other hand, tend to have a very hard time with this sort of instructions because they need to load new samples every time an important parameter (like location, radiation, projection) changes, and that requires extra CPU-effort and time (which is why the pre-Hyper plugins from Acustica Audio were such depressingly sluggish and resource-hungry things. Acustica's Silver, an excellent sounding convolution reverb, and the same company's Ebony reverb (also fantastic sounding) are, to this day, still barely usable in a real-time DAW context because even a far-above-average computer can't deal with these plugins' resource demands).

But I’m sure a good versatile and practical solution — by ‘versatile’ I mean: not limited to one specific room — is not too far off. Someone somewhere must be working on this. I can’t imagine it being beyond today’s (and certainly not tomorrow’s) software, to create models and profiles for all the instruments (and their variants), for a comprehensive set of fully customizable rooms and for the various ways in which various sources bloom in whichever environment of choice (including the simulation of all the details related to location, distance, projection and reflections), and provide all that data in a convenient spatializer / bloom generator.

__

[Guy Rowland]

Apparently you can't move things on playback, so that settles that one.

In theory there's nothing special about the tech of this plugin (although it's FAR more impulses than normal), but it does sound particularly good imo.

I agree with your last para. I'm surprised there's not been more R&D on it - I will forever think that vast mult-mic libraries are a wildly inefficient way of going about things.

[Lawrence]

VI-C thread now at 19 pages-just sayin’.

[Luciano Storti]

VI-C thread now at 19 pages-just sayin’.

Perhaps not as relevant of a metric, simply because it's about a Spitfire product? Having said that, I am liking the sound more and more.

[Guy Rowland]

VI-C thread now at 19 pages-just sayin’.

I did read someone say there that if it doesn't hit 1,000 it'll be a disappointment. Spitfire, Air and Reverb is peak V-C.

[Geoff Grace]

In the meantime, the Gearspace thread has yet to reach page two. Most of those who posted there seemed to reject it because of the lack of demo combined with a high price.

Best,

Geoff

[Lawrence]

At this price point, a trial period would be nice.

[JohannesR]

I'm still attracted to the model of basic close mic set up, using fx for the room.

So I'm going to be following this closely. CPU and real world experiences are particularly interesting to me.

Real world experience; Alan Meyerson tend to leave the close mics out of the reverb sends, which aligns with my experience and taste as well.

The close mics doesn’t drive reverbs well. It adds some sizzle, and doesn’t contribute to a nice sounding and believable acoustic space - at least in my ears. Heck, on some libraries (particularly ones recorded in smaller spaces), I wish there was less close mic information - even after muting the close mics. To each their own, I guess

[Ashermusic]

I don't know if this will jibe with anyone else's experience, but I have had the chance to record in some really legendary studios here in Los Angeles, great sounding rooms. But in the same room, with different engineers, different musical material, different musicians, it's never been the exact same sound to mix. So the room, good as it may be, is only one ingredient in the cake.

I have a ton of great software reverbs to choose from. Until A.I .really does start to play a significant role interactively with a software reverb I simply can't see myself buying another, regardless of how storied the venue is.

[Scoredog]

I am in agreement with Jay.

[JohannesR]

I agree about recording in LA, @ashermusic. But have you ever been in Lyndhurst Hall? It is a very special place that doesn’t quite sound like anything else. It is the one studio I think it actually makes sense to name drop when selling sample libraries. So I completely understand the rationale behind this plug-in.

That being said, I haven’t bought it either

[wst3]

of the rooms where libraries have been recorded my favorites, in no order whatsoever, would be Project Sam, Orchestral Tools, Spitfire, and Cinesamples. No two alike!

Can any of these rooms be accurately reproduced in software? Doubtful today, but perhaps in the future?

Can any of these rooms be "modeled" in such a way as to provide the same general feel - that part that makes us sit up an notice?

I think so, and I think the Spitfire offering might do just that. I've never been in the room, so I can't say for certain, but I like the feel that the plugin adds -- of the demos I think the french horns were my favorite, but I used to play horn, so that might influence me?

The sound of Air Lyndhurst is not a sound that I can imagine using in any forthcoming projects, which tend to lean towards Ocean Way. If I had a project that would sit nicely in Lyndhurst hall I would not hesitate to purchase a license. Not even for a second. But until that time I am with Jay on this one - I already have far too many reverbs!

[bbunker]

Didn't read the whole thread, thought I'd check it out and JUMPING JEHOSOPHAT it's basically $300. Close window.

Just wanted to clarify something: 'modelling,' 'convolution' and 'algorithmic' are a bit misused in this thread. In recent history, most of those 'room-authentic' reverbs haven't really been convolution reverbs in the sense that old-school convolution reverbs would simply run a 'one-dimensional' white noise tensor into a convolution network and get out a 'two-dimensional' tensor that extends into the time field, but definitely make use of convolution networks, and much of the difference is largely semantic. Any convolution network is really a form of Machine Learning, so feeding additional data to it will result in better results - after all, what makes single-shot convolution impulses lacking is that they are profoundly overfitted. Are IK, Softube, or UAD feeding additional single-frame one-shots into that input tensor? Are they feeding in additional information (i.e., weights or biases) to make it a 'directed' network? Are they using 'two-dimensional' tensors based on actual instrument performance, in the same way that IK's Tonex hopes that any overfitting will be in the direction of things that sound good? Don't know - they're not saying. But any of those modelling reverbs are definitely not algorithmic in the sense of Schroeder, Gerzon, etc.

I wouldn't be surprised if we see an old-school convolution reverb tagged as being an "AI" reverb, soon - because a simple, one-layer CNN is a (rudimentary) form of machine learning, and we've decided as a collective that ML is AI. Or maybe they'll start doing these things and be upfront about it. Spitfire could have, for example, assigned numerical tags to the 'room baffling' settings, fed that along with the input tensors, to have a control that would move linearly from no baffling to maximum. Or room position itself could be a two-dimensional tensor to add to the input. You'd need a lot of data, but if you shoot 67,000 impulses, you've got data. My guess would be that's what we'll see - after the first scammy 'this was AI the whole time' reverbs, of course.