Get the sound of top-tier studio microphones with LAVA from Acustica Audio
by AcusticaCM   8th September 2022

I've used similar products in the past from Antares and IK Multimedia. Neither could turn water into wine like this technology claims. To me this is due to issues like "capture" and "resolution" which cannot be replicated with EQ/compression/saturation/convolution etc.

I now use Slate VMS which I'm very happy with but it is a closed system with a very detailed and neutral source mic and preamp. In the same way I prefer Slate VSX over Sienna, but of course like everyone else on here I'm happy to be vowed (and proved wrong!) by new and forward-thinking tech!

I hope we can surprise you! there are many more features that we will leave for the post when it is released. Stay tuned.

You can't do that, but we are conducting an interesting experiment during these hours, and we will publish extensive results on that.
Join in, you'll be surprised about the results.


We recorded an acoustic session (the one in the clips above) with a rather inexpensive microphone, a 57. We chose, at the outset, the budget microphone example, the one most common among home studios, the one that people would gladly trade for a more expensive and famous microphone. The 57, along with the 58 (which ironically produces a better result with Lava) is an extremely popular microphone; virtually everyone knows it.

Moreover, we are already talking about an extremely complicated task since most people need a microphone for vocals and nothing more (in general, a recording studio that needs to mic instruments already has a fairly assorted set of microphones, and does not need mic modeling technology, with a few exceptions).

We simultaneously recorded with a relatively decent microphone, costing 27 times more, a u87. Why this particular mic? Because it's a popular mic here in our neck of the woods, it's one of the "goto" mics for vocals, and most importantly, many people are pretty knowledgeable about its sound because they often have one at home.

This is a typical task, and we would expect from a critical audience like our audience the ability to distinguish between a 100 euro mic and a 3000 euro mic, even considering the ultimate complexity of the task (not only vocals but also a miked acoustic guitar).
Hearing these two tracks, it would be easy to distinguish the one made with the better microphone. But we applied Lava and must admit that they sound quite similar.

Are the two clips identical? I would say no. But here's where the fun comes: 1533 users have responded so far, a remarkable number. We have reached a user base of professionals, not the end users, who take the tracks from Apple music and listen to the Dolby Atmos binaural mix in the car Bluetooth because they can't tell the difference between a binaural mix and a traditional stereo.

And we have an interesting result:
the correct answers are 795
the wrong answers are 738
This roughly corresponds to a random rate; we consider pure randomness at ratio 1; here, we have 0.928. It shows unequivocally that in the comparison, people with skill cannot tell which source was recorded with a microphone that costs several orders of magnitude more.
And so, as you say, you can't turn water into wine, but you can start with water and disguise it enough to be comparable to wine in direct comparison, depending on the circumstances, to the point where people don't know which drink is from wine and which is from water. Not all microphones are suitable for this transformation; for example, the same task with a Behring mic produces a fairly good result for vocals but insufficient for guitar.

However, it should be understood that such a tool, when it works, solves a problem. If "experienced" people cannot distinguish the two recordings by direct comparison,
- what could he understand if he did not have a reference? here we posted two clips and gave you plenty of time to think about it. If it had been a single clip, how would you have reacted?
- and then why not go further and apply a really rare and expensive microphone? if we can fool an outside listener with a "difficult" task like the one described so far, let alone what happens when you apply a rare, vintage microphone that few people in the world have the ability to distinguish from cheaper ones.
The task becomes turning water into vintage cognac-how many people know cognac?


We will provide a way to be able to check these results.

clip 2 sounds like a condenser microphone

Well now, it is not unequivocal at all really, is it? It's quite up for interpretation. You are assuming an awful lot about the people responding to the challenge.


Does it mean that half of the people with skill can't tell which source was which (as you are suggesting)?

- or -

Does it mean that half of the people aren't skilled enough to tell the difference?

- or -

Does it mean that half of the people don't know which attributes to listen for?


I'm very interested. I am supplied with my fair share of sources that were not recorded well and that could use convincing tech that actually works. I think it would probably be best to stick with reporting the numbers and maybe not drawing too many conclusions. Your users will draw their own conclusions without you needing to make such statements.

You left out an option, 2x half the people can't hear significant differences because its a lossy Soundcloud compressed mp3 file (unless you subscribe as a LISTENER as well as upload a high content file)...I would venture to say that o the 1500 or so votes probably zero pay for the Soundcloud Go +...

I could be wrong of course but there's a huge difference in listening to files in a treated room with high definition and mix/master level speakers versus listening on a laptop or tablet with 1" speakers built into a case...

Its the youtube/soundcloud enigma of why my mixes don't sound like Bob Clearmountain or Greg Wells...I pushed all the right buttons according to Puremix...it's gotta be something between the keyboard/mouse and chair I'm sitting in.

It is quite a representative sample because it involves a diverse international audience with above-normal skills.
A not insignificant portion of the sample owns one of two microphones, some both.
The number and the result already achieved now is already interesting, and it already represents an obvious way a nearly random distribution.
We can stand here and argue all we want, but even if some people barricade themselves behind a "they sound completely different," the truth is that it does not seem easy to select the origin of a microphone that costs twenty-six times more than the other while having time and both tracks available.
It is easy to downplay the result, but the use for guitar and vocals is why the investment of a major microphone is made; if even an expert cannot tell the difference, I wonder what the end audience will understand.
Remember my words said today when someone in the future argues that "unequivocally our plug-in cannot turn water into wine and that there is a gulf with the real thing or that this approach serves little or no purpose and that it is better to record the sources well because there is no going back."

Btw when you apply for the competition you'll find the full quality wav audio files as well

There is also a couple of full quality wav file in that page (the competition landing page), and people are using it

For microphone modeling to be done successfully, it's necessary to emulate several different things.

• On-axis frequency response. This is easiest to model, at least in broad strokes. You can do something similar with one of those equalizer plug-ins that can match a spectral sample.
  
  Where this falls apart is in dealing with narrowband resonances and with nulls caused by discrete reflections. A resonant peak (in either the reference or recording microphone) is likely to vary in position slighly between one serial number and another. Attempting to compensate a narrow-band phenomenon when you're not entirely certain where it lies is likely to make things worse rather than better. Furthermore, a frequency null due to an internal reflection can't be fixed even if you know its exact frequency. (A "zero-forcing" equalizer is a theoretical construct that cannot exist in the real world.)
  
  
• Microphone nonlinearities. One of the reasons we value valve mics for recording vocals is the way they compress and smooth things out and how they add interesting harmonic content when doing so. A lot of what people like about certain dynamic microphones on snare drum is the way their transformers saturate.
  
  Here, Acustica Audio brings to the table a high level of expertise in modeling non-linearities developed for their emulations of rack-mount gear. They are probably better at this than the makers of existing modeling mics. Now it's one thing to add distortions, but quite another thing to remove them. The only people having iany success in the latter are certain top-tier makers of loudspeakers and horns. That's because they manufacture, with tight control, the transducers they are trying to correct.
  
  
• Polar pattern replication. We often choose different mic patterns based on how they differentiate direct sound from room sound, how they control leakage, and what that leakage sounds like.
  
  Most currently-marketed "modeling" microphones attempt mimic these effects by employing a dual-diaphragm capsule and equalizing each side seperately. This is a first-order solution to a high-order problem. Off-axis colorations that exists only at certain angles can't be fixed, even with dual diaphragm EQ, because first-order pattern synthesis is insufficient to model these narrow-angle phenomena.
  
  Acustica's claim that they can work with any microphone (most mics are of single-diaphragm type) means that they can't do pattern emulation at all.

David L. Rick

This type of technology simply cannot deliver what it promises, for reasons that have been pointed out umpteen times by knowledgable people who work with real microphones daily for a living. Everything from polar pattern to off-axis response to differences in sensitivity make claims of 'accuracy' simple smoke and mirrors and marketing mumbo jumbo. I honestly find it a bit stunning that companies continue in their efforts to market this as any sort of professional audio solution. It devalues those brands by making them appear as unserious at best, or simply cynical and even predatory at worst, IMHO.

YMMV, etc.

exactly as with the plugin, there is a very simple macroscopic result: instead of imitating the flow of electrons down to the last quantum of matter, general features can be imitated.
In the end of an instrument it matters little if it behaves identically to another in every possible use, this would be impossible and we agree.
What matters is the end result, that is, what the instrument produces and what a simulation system would have produced, and whether the task assigned at the end is sufficient to pass a validation test of some kind.
How many people who buy an 87 really want to play with polar patterns? some people get a microphone to sing into it, to have a "kick-ass" track and drown it with autotune and are interested in producing something that you don't believe a negative criticism from the people on the job, who are always ready to say that you have to sing the right mics, with the right converters, with the right listening, in the right room because that's the way to do things.

In doing so you sometimes lose one of the main advantages, which is to access some of the nuances that we couldn't afford even in our dreams, such as "imitating" the behavior of a particularly expensive microphone, and having a magic that we wouldn't know how to recreate otherwise.
One of the most interesting uses is not to take this 57 to an 87, but to a 251. Again, we will publish the final track at the end of the contest, so as not to pollute it with new data that, cross-referenced, may influence the answers

Yes, your thought was somewhat my thought. We didn't pretend to "replace" a mid- to high-end microphone with a low-end microphone.
We were thinking more along the lines of headphone correction: bring the single headphone to higher performance, change some unpleasant characteristics, equalize them to a higher level.
To this day I remain of this idea: some microphones cannot imitate other microphones in all possible uses, due to a number of mechanical limitations. This is also true for the "ideal" microphones provided by some companies: they may be neutral and very well known and predictable for some characteristics and may have an easier correction (e.g., they are all equal to each other so the correction curve is constant and works well), but they cannot physically become a different object.
In some cases the task is impossible: for example, some microphones cannot imitate others perfectly on complex tasks such as acoustic guitar or piano.
However, there are sometimes simpler tasks (such as voice) where it is easier to fool the listener, and this was our starting point in developing the product.
Along the way we marveled at how some microphones could replace others for complex tasks, where by complex task I mean miking an acoustic instrument with a very wide bandwidth (because there are harmonics, for example).


The idea for this "competition" came to me one day when Piero Sturla, a consultant who helped us find many microphones, worked actively on the project, and whom I consider one of the greatest experts in italy on microphones and comparison methods, as well as field experience, arrived in the office.
He had never heard the plugin in action, and he confused just a plugin-corrected sm57 with a u87 on the first try and the first listen.
I thought a very simple thing, which is that if he could have been fooled, then probably a good part of the expert community could have been fooled, and probably the entirety of the end-user community (the streaming platform users) could have been fooled.
He confused the mics by being able to hear both versions, calmly and with the precision of a pair of headphones he knew (his own), on high-quality audio and with a past behind him that boasts comparisons of hundreds or thousands of different brand mics, choice of capsules for various projects, work experience with a rather world-renowned Italian boutique brand, and years of focused listening in pursuit of perfection.
Certainly he heard the different tracks, certainly he was able to recognize on the fly which of the two tracks he had heard previously and in what order, but he was not able to understand which was the real microphone and which was the simulated one.
This happens because our ability to memorize a sound or timbre is extremely impaired by our status as human beings, and it is a matter of memory, which of course some people have in abundant quantities, but let's not kid ourselves, most people who are not educated in listening don't even know what it is.

You could extend this kind of contest to software comparisons with mystical instrument hardware, equalizers and processing units, and we would probably arrive at a similar result. I think in that case getting closer is critical, because when I take a track and process it with a mastering chain where plugins are loaded dozens or hundreds of times, a small difference reverberates into a gigantic difference.
However, this is not the case with microphones, which are practically applied once to the source, usually dealing with extremely frequency-limited input sources, with a fair amount of noise floor, with limited dynamics (and to be limited further later), and with rather unique and not very repeatable performances, so sometimes you can't understand what contribution the capturing instrument brought and what contribution the musician brought with the imperfection of the performance.

And here we come back to my point: how many times have I heard a singer asked what microphone he used in a track and to the answer that indicated a cheap microphone I heard "you might want to do it again with a proper instrument."
This is not true if you have the right correction tools.

In the coming months you will hear recordings that may have been helped by Lava, you will be constantly questioning what was used, because you will know that there is a plugin that can fool a rather experienced audience by turning water into wine in some cases.

This is gonna be very interesting to the say the least!! I used to own the Slate Digital VMS system and it was a bit of a hit or miss. Some of the emulations sounded good, some of them just sounded really poor -- though it may come down to my own personal voice. I currently am using two mics now: a Neumann TLM103 and a Sennheiser MKH416 which I both use for voice-over. I don't intend on passing them up, but I'm always looking at the latest in technology.

Honestly, as someone who's played with many microphones, but is neither wealthy or privileged enough to sample every color or every mic I want, I'm always interested in forward-thinking tech. Imagine being able to have this massive virtual locker where you can pick and choose the perfect mic for every singer and song. Slate Digital did a great job moving things forward, so it'll be fun to see how AA takes it further!

The most obvious glaring fact that often gets overlooked is that regardless of IR vectors, or EQ matching or whatever the latest fad is you cannot treat what you have not captured...

Those expensive tube mics capture frequencies based on capsule design combined with a bunch of other technical factors and ALL mics are bandwidth limited by design...

Now you can add all kinds of EQ sweeteners to the mix but like I suggested you cannot treat something that the original mic did not capture: the dynamic range, the sensitivity of the transducer, the THD, and most of all the frequency response whether on axis or off.

I'm all for advancing technology and as a general rule I like a lot of Acusticas stuff in spite of GUI's from hell...and I'm sure this will add some toasty sweeteners, but physics is physics.

I think this does indeed boil down to whether it "fools" listeners, and to zaphod's point...

This contest made me think of truffle oil. Being able to emulate a genuine truffle aroma in a extract or oil is a holy grail.

If you can produce a artificial truffle oil that captures enough of the qualitative and quantitative aspects of a real truffle that it confounds reputable diners and chefs then you have a successful product. End of story.

It doesn't matter if you use a different sulfurous thiol than is found in nature. You still win if you convince most of the people most of the time even if you don't convince all the people all the time.

Of course, this is a "contest" and not a scientific study, but the idea does indeed lend itself to running blind tests on defined levels of skillsets.

The two samples were veeeery different, but I don't have enough first hand experience to know which is which. The clip 1 has a much lower frequency content and some midrange goodness (but is a bit muddy) while clip 2 has nicer highs (but is a tad thin and reedy).

The SM57 doesn't have the specs to get as low in the frequency spectrum as clip 1 when viewed by an analyzer.

But then I'm wondering if that low bump is being generated by the LAVA tech rather than the U87 since the U87 shouldn't have have the upper freq limitations apparent in clip 1.

Just submitted my answer, can't wait to find out!

sure... But people are not choosing the right clip

Hence my previous post about lossy audio clips on Soundslike Cloud...and for good measure I wonder how many of the people voting actually use high end tube mics on a regular basis?

I doubt very many...I can pick out our Townsend, or Slate or Antelope mics when compared to our tube mics almost every time...we never drag those out anymore.

Of course there are a ton o factors as to why you could need a mic emulation...and in those cases (source material, budget/availability/etc) an emulation is perfectly acceptable...

I can't tell which is why since I have no real experience with either, except that I prefer Clip B. If that's the sound of Lava, I'm a happy camper

I prefer Clip B

Hi, contest is over according to site. Could you please let us know how long it might be until you post the correct answer and results? Thanks!

Yes, the contest is just over; tomorrow, after the release or Wednesday (tops), there will be an announcement.

I just noticed that you are the thread starter for Acustica general thread. Thank you! So much great consolidated info there.

Wonderful, thanks. The anticipation is killing me. Just saw you sent our loyalty gift discounts with Lava included so we are all set :-)

Acustica is pleased to present Lava, the microphone revolution plug-in. A winning choice for recording, mixing, and broadcasting applications.

What if you can get instant access to the best microphones in the world in one simple click?

Lava is a VST2/VST3/AAX/AU Acqua plug-in that accurately recreates the tone of classic microphones, including the microphones' response, harmonic content, and proximity effect. It is an out-of-the-box plug-in that can make your own microphones sound just like the microphones you have always wanted to own, with a few simple clicks, and you can use it on any vocal and instrument. You can also use Lava in real-time by using the LAVA ZL version (its latency at 44Khz is 4ms.)

No matter what physical microphone you have, just select it from the menu and make it sound like the one you want by choosing it from the list. Done!

Compared to our other plugins, Lava is particularly lightweight regarding CPU consumption, and we are sure you will appreciate it. Plus, phase responses are nice and clean and without artifacts.








With Lava, you can easily transform your microphone by choosing between a wide selection and making it sound like the one you desire, transforming a $100 microphone into a $9000 one.


Main features

• Over 140 microphone emulations.
• Distance slider to remove any proximity effects on the source mic.
• Dry/Wet Mix knob.
• High-pass filter from 45 Hz to 1k1 Hz.
• Input and output led meters.
• Low CPU usage.
• ZL (zero-latency) version suitable for tracking.
• Up to 4x oversampling.
• Resizable interface*

*Choose between 3 magnification values (1x - 1.5x - 2x) from the top right ▼SIZE drop-down menu.
Once the desired size has been selected, the plugin must be removed and re-loaded to apply the changes. This action affects the currently selected plugin. New instances of the same plugin will open with this magnification.


What you get

Lava includes over 140 microphone emulations (100 source and 41 destination mics) for a total value of $170k worth of microphones right in your DAW.
Instantly expand your collection with vintage and modern microphones models on the market, including a selection of ultra-rare boutique mics.






Note: as we have experienced, if you are looking for maximum accuracy, some conversions physically cannot be done (from a Figure-8 mic to a hyper-cardioid one), but in the everyday practical use of a mic, the approximation we make is enough not to distinguish which mic has been processed with Lava and which is the original one. You can barely perceive the difference (as proved by the contest!)

Tip! we recommend trying the EM-251 (ELA M 251) emulation as the output microphone as it sounds amazing after processing, and it is a rare and expensive piece of engineering (but of course, the choice is yours, depending on the instrument or vocal you need to process).


Lava - Demo





The story behind Lava

Lava is a project that took over six months of development to properly understand how best to capture the sound characteristics of each different sampled microphone.

We started by choosing the most suitable environment for the measurements: an anechoic room. In particular, we used the one inside the 'Politecnico di Milano' run by Professor Roberto Fumagalli, who spared no effort in his helpfulness and precious advice during our extended stay.
This place also has a complete absence of reverberation in high, medium, and low frequencies meaning no room modes can change the sampling transducers' timbre.

Once the ideal location was found, we moved on to the choice of the speaker, the transducer responsible for emitting our proprietary measurement signals. After several attempts, the choice fell on passive Rasch Audio© speakers by Simone Fagnani, driven by a hi-fi amplifier.
We were particularly convinced by the phase coherence and the very low impulse response, essential characteristics for not emphasizing data capture. The 3-way speaker with its dedicated sub was measured with special calibration microphones and a calibration file. This way, we could cancel out even the smallest emission dissimilarities.

The measurements were carried out under controlled temperature and humidity thanks to the instruments in the chamber. Laser levels and laser meters helped us to achieve repeatability of positions and operating distances.
Once the measurement routine was established, we moved on to find a conspicuous number of microphones that represented the timbre, engineering, and history of these fascinating transducers. We were helped in this task by Piero Sturla, a technician and microphone manufacturer for Ribera Audio, who is well known and appreciated among professionals.

Piero ensured that the electronic functioning of every single sampled microphone was perfect, especially for the vintage ones, devices that now have more than 70 years on their shoulders. In fact, each microphone was carefully checked in its operation before and after measurement, verifying the data acquired with those released by the manufacturer.
The pre-amplifier used for the acquisition is among the best in terms of linearity and driving capacity, its generous input impedance being able to handle even the hardest microphones. The converter used for recording and testing signal output is a benchmark for its linearity and absence of harmonic distortion.
A heartfelt thank you goes to Marco Vannucci, Acustica's collaborator, who actively participated in the project and contributed significantly to its realization.



Technical information

• Plug-in Format(s): VST2, VST3, AAX, AU
• System Requirements
• Available sample rates: 44.1 - 48 - 88.2 - 96 kHz
• Windows 10 compatible
• Apple Silicon M1 macOS Monterey compatible



Price and availability

Introductory price: €129 | 38% off until October 11, 2022 (reg. price €209) Buy Now!


The product is at an introductory period for the first four weeks after release or until the first official version is released; this implies that the product may receive improvements, changes, or fixes available through Aquarius Desktop updates during that time.


How to install the free 30-day trial
Note: Lava trial version includes 44.1 - 48 - 88.2 - 96 kHz sample rates.
Please be careful and remember that trials expire 30 days after authorization, and we strongly recommend not using trial products in commercial sessions or any important project.


Lava user's manual


Lava web page: https://www.acustica-audio.com/store/products/lava

Is your microphone not on the list? You can now request it through this form: https://lava.acustica-audio.com

And we are out! https://gearspace.com/board/new-product-alert-2-older-threads/1388140-get-sound-top-tier-studio-microphones-lava-acustica-audio-post16164359.html

Been using this for couple of weeks and it is an amazing tone shaper.
My favorite uses have been overheads, drum rooms, snares and vocals. It really can change the timbre and character of the source surprisingly well and because of that I have been using less other processing.

For overheads - try Flea M49 or Neumann M149 as the target. I have character wise turned some OH recorded with vintage U87 into m49 quite well. The vintage u87 were one of the only ones available at that studio and they were not the best fit.

For vocals it is a choice of both of the before mentioned ones plus the ELA mics, C12, u47. The u67 has never worked for me, apart from that one time when I needed a dark mic. The distance slider is useful to dial in the upper midrange bite and harmonics.

For snare try converting your sm57 into that MD421 vintage if you are looking for a big punch in the chest. Or try different flavors of sm57 as they are different.

For kicks - I had good luck turning a u87 into u47 couple of times.

I am not saying that this can replace having a large arsenal of quality mics but this gives a way to give some of your mics a different voicing during mix time.

Also I have tried to replicate the EQ curve of lava plugin with a digital eq but with no luck. The digital eq'ed version still sounded like an eq'ed track where the LAVA processed tracks exhibited a different tonality altogether.

PS just now converted u87 into an ela 251... wooh, the presence is remarkable.

PSS sometimes some mics show up some weird resonances up high that sound not great but it is possible to notch them out with a digital eq and the result can still be better suited to the track than the original.

I'm so buying this!!!

Would be nice if they included Slate mics as a source, those sound so good and sterile (in a good way) that the Acustica spice&treatment would be superb (no doubt about it)!!!

My source mics that are mostly collecting dust:
Audio-Technica AT 2020
Behringer B5
Rode NT1 A

Edit: Bought, without trying

Going to try this right now, would respectfully request that 48k be included in all future demos. 88 is way less common and I see why leave that one out but 48k is so common that I firmly believe its a worthwhile inclusion. Fortunately the track I'm currently working on is 96k so all good for now

I must say that the list of source mics is a big let down.
For instance, there is not a single dpa mic.
I hope this list will get a lot longer, without having to purchase additional packs.