Someone said this was Snake Oil...

[Ploki]

All this mumbo jumbo would go away if people understood better the limitations of human hearing and the implications of the Nyquist theorem, but here we are in the year 2019 with flat earthers, anti vaxxers, "DAW sounds better than other DAW's" and people paying several thousands for converters.

It must be true that most artists suck at math.

This would be funny if I was observing it from the moon

[jochicago]

We all need to reach out for guidance for things we don't yet fully understand. We need opinions/reviews and recommendations.

A big problem is taking a brand's word for it. Every emulators is the best, every eq the cleanest, every compressor has the most mojo. If you are still listening to a brand's own sales pitch, do yourself a favor and stop. Find other means to determine the worth of things.

This applies to anything we buy today, particularly anything promoted by American brands. Not to diss a particular country, but here it seems that we abandoned reason years ago. It's all fluff coming from every angle, food, tech, clothing, news, you name it - nearly all of it fluff. We have already abandoned even simple objectivity in news and facts. We are now starting to abandon even basic science. I just saw an estimate that about 6 million Americans believe the earth is flat. That group thinks they are on to something powerful, they need to spread the word and "enlighten" others with their new found wholesale disregard for logic and the simplest of evidence.

And that's why I hate Apple. (To bring this back to the current thread topic). It is one of the precursors of the movement "you don't need any facts or info when you have me to tell you what to think". It was cute for a few years, but now that the culture has become mainstream we need to revolt against this nonsense. People standing in long lines to fork out top money for a device they can't afford nor need, that is at least a year behind the technology curve. There's nothing to be gained by reverting to the dark ages, yet that's where we are currently headed, proud in fact-free stubbornness, self-righteous, with many chanting something about a wall.

Next topic: let's address Beats by Dre.

[tomheavybeats]

Drivel

[jens]

See? I didn't mention my music with any word nor did you listen to any of it. Yet all you got from what I wrote was that you're the better musician because I "make music for scientists".

You'll be grasping at any last straw in order to feel better about yourself. That's how they got you.

[imrae]

Example: IR reverbs are not perfect, in fact most reverb experts would call them so insufficient as to be labeled useless. IRs on their own decay and can't bounce around, when many top reverb algos can swell, bloom, ping-pong, etc. A sampled IR without the algo parts is just 25% of the required tech.

This is nonsense. IR is not "useless" it is used widely; it does let audio "bounce around" a space; it is trivial to construct an IR for ping-pong; there are IRs out there for the famous midiverb "bloom"; swell is just a volume envelope on the IR which can be baked into the IR if desired.

Methinks you've been taken in by some fake "reverb experts"?

[jens]

Example: IR reverbs are not perfect, in fact most reverb experts would call them so insufficient as to be labeled useless. IRs on their own decay and can't bounce around, when many top reverb algos can swell, bloom, ping-pong, etc. A sampled IR without the algo parts is just 25% of the required tech.

This is nonsense. IR is not "useless" it is used widely; it does let audio "bounce around" a space; it is trivial to construct an IR for ping-pong; there are IRs out there for the famous midiverb "bloom"; swell is just a volume envelope on the IR which can be baked into the IR if desired.

Methinks you've been taken in by some fake "reverb experts"?

There's been a lot of good explanations from real reverb experts as to the shortcomings of IRs.

Quite a good one of some of them is here:

http://www.quantec.com/index.php?id=about_irs&L=0

http://www.quantec.com/index.php?id=faq041


Also Michael Carnes of Lexicon and Exponential Audio fame explained some of it here:

https://www.gearslutz.com/board/high-en ... ary-3.html

Convolution can be very nice and is a useful tool for things other than reverb. But it always gives you a fixed impulse pattern recorded from point source(s) to point destination(s). Real reverberation is a little less linear and real performers rarely cluster in a fixed point. In making a mix, you're generally interested in getting the right sound for a particular instrument or group. An algorithmic reverb is easy to adjust. With a convolver, you've got to go looking for another impulse response.

to which Sean Costello of Valhalla Audio added:

You can redesign an impulse response on the fly if it is synthesized, or you can apply amplitude and filter envelopes to an existing response, resample the response, etc. However, it will always be time-invariant, which is not the case for a real room or hall. Plus, convolution is a resource hog compared to an algorithmic reverb running on most machines. I have seen a time-varying convolution algorithm, that cross-fades between different inpulse responses, which seems completely wasteful from a resource perspective. Using 70% of a modern CPU to emulate a reverb that originally ran on a 6 MHz part...

And again Michael Carnes:

I guess you'll remain a little bewildered then. A convolver is no more a representation of a real space than an algorithmic reverb is. It's an accurate represention of an impulse reponse from a speaker or two to a single listener position. As I said before, a real orchestra is spread out in space. So is the audience. The impulse response from the second clarinet to seat 7C is different than the impulse response from the first clarinet to the same seat. A convolver effectively squishes the entire orchestra into a couple of points, quite accurately mimicking descent into a black hole.

On top of all that, you simply don't hear every single reflection in a hall. That's not the way the auditory cortex works. You may hear discrete reflections for a few tens of milliseconds, but after that, reverberation becomes a frequency effect (with interaural phase differences being quite important). If you do that part well, the listener is satisfied.

I didn't say that convolvers were bad. Sometimes they sound very nice. But to somehow say that they are natural and algorithmic reverbs are artificial shows a little too much susceptibility to marketing. News flash: They are both artificial.

[imrae]

I don't disagree with those people; these are the real limitations of IR. The linearity, cpu cost, time-invariance, limited choice of sound sources...

The post I quoted mentioned none of these things and instead listed a bunch of stuff IR can in fact do, while falsely suggesting that experts consider IR to be "useless".

[jens]

Fair enough.

[jochicago]

Example: IR reverbs are not perfect, in fact most reverb experts would call them so insufficient as to be labeled useless. IRs on their own decay and can't bounce around, when many top reverb algos can swell, bloom, ping-pong, etc. A sampled IR without the algo parts is just 25% of the required tech.

This is nonsense. IR is not "useless" it is used widely; it does let audio "bounce around" a space; it is trivial to construct an IR for ping-pong; there are IRs out there for the famous midiverb "bloom"; swell is just a volume envelope on the IR which can be baked into the IR if desired.

Methinks you've been taken in by some fake "reverb experts"?

Let me elaborate so that my meaning is more clear.
IR lets audio bounce around - Not it doesn't, not in a natural way. In a flat way. The point is that it doesn't naturally replicate a room, any more naturally than an algo reverb does. In fact, algos can do a better job of representing a 3d space because they can mimic instead of playing a flattened sample.

IRs can bloom - Any convolution player introducing a swell or bloom is using an algo to do it. The IR itself can't swell, it only decays since it is a sample of an impulse in a room, it decays the whole time.

I'm not saying that every IR reverb plugin out there sucks and can't do these things. I'm saying that these things are done with algos. You may start with an IR, but that's not all that's going on in a bloom reverb.

If you look up the Lexicon free IRs, somewhere there's a quote from Lexicon saying that they are ok with people taking IRs of their machine because basically you are shooting yourself in the foot if you think you are making a comparable reverb with just an IR. Their box is doing so much to move the image in space, chorusing, swells, etc, none of that will appear in your flat IR image.

And that's what I meant about experts thinking IR-only reverbs are not right (insufficient, useless on its own), because you'd be missing a lot of what we can do to a reverb with algos to make them more interesting. Without adding some logic to move the IR in space, swell, etc, if you only play the original IR in a convolver you only have a flat-ish decaying room.

In other words, if someone like Valhalla wanted to make a reverb that uses IR, you can bet your house they'll surround it with algo processing to make it work. There is no point whatsoever of making an IR-only reverb when all the power to command them and make them interesting is in the algos. That's why the best reverbs are mostly algo stuff.

[jens]

While talking about emulations and impulse responses:

in the case of reverb, just as convolution reverbs, no algorithmic will ever be able to really emulate a room. Before we mentioned that every digital emulation by its very nature must be an approximation. But I think in the case of emulating electrical devices, emulations can get a lot closer to the original as compared to reverb, if they are spice-based (or using a similar component based approach)as they can emulate the whole circuit itself. With reverb that would mean emulating every inch of the surface of the room and also every stationary and every moving object in the room - everything. This is of course not possible (it might be one day).

Convolution - and that includes the dynamic one Nebula applies - is always limited to taking snapshots of whatever generates the impulse responses. And how many of these snapshots you may ever take*, it will always be only snapshots of the process-result what it seeks to emulate, never the thing itself. And hence I think it can be argued that in the case of emulating electrical devices component-modelling can approximate the device a lot more closely than dynamic evolution ever could, since it - as I mentioned before, is able to emulate the circuit itself and not only what comes out of it. So, at least in theory, the emulations of Fuse Audio, Cytomic, Overloud and others should be able to emulate the devices a lot more authentically than Acustica Audio ever could.

Hence I would assume the Nebula-based emulations - despite using a lot more ressources (CPU, RAM, disk-Space), having a much less responsive user-interface, needing to re-load samples for certain parameter-changes and being prone to extremely weird and even show-stopping artifacts - to be less accurate and subsequently much less perfect emulations of the analog reality it seeks to emulate.

In other words: I would expect Nebula-based stuff to sound much worse than the best spice-based emulations.

(and whenever I tried some of their stuff or listened to some examples as in the video of the OP here in this thread, my ears told me the same - but that of course was just my subjective impression and very well could be extremely biased - but to me personally Nebula-stuff tends to sound bland, boring, dull)


* https://youtu.be/VNftf5qLpiA



Spice-based emulations are only limited by their internal sample-rate. What's the snapshot-limit of Nebula?

[bmanic]

exactly this ^

for every developer who claims doing accurate emulations of audio hardware
be that may u-he, slate, roland, uad, tal, etc etc there are many side-by-side
public comparisons of the hardware and software set to match as perfectly as possible.

EXCEPT acustica, where the exactness of the emulation is an article of
religious dogma, and side by side comparisons are NEVER presented either
by the company or the fans.

So if you like acustica, that is charming, but please do not profess that it is an acurate
hardware emulation until you produce the proof.

It's interesting to keep hearing this lie being thrown out there over and over again. First of all, it is by far a MINORITY of developers who have done proper A/B comparisons between hardware and software. You make it sound like it's the norm. It is not.. not even close.

Also, Acustica audio has at least one product where they did an exact A/B comparison video on their youtube channel as so many people were complaining, especially when the hardware in question is bespoke and not available for purchase.. so they did one. Yet conveniently nobody mentions this.. also all of the naysayers interestingly completely shut up once they did that but have apparently started to resurface yet again.

In case you want to go check out that A/B test it's their Cobalt product and the 2nd half of that introductory video of the actual hardware contains exact A/B comparisons.

There's also been a number of shootouts with hardware vs software where Acustica Audio plugins or 3rd party developed programs have taken part.. so it's not like it's impossible to find any comparisons.

Still, the exact same caveat remains as with any other plugin. Does it _really matter_ if it's exactly accurate with hardware? Why not ask if the plugin is useful and sounds good? YISH313 is at least providing extremely easy to follow before/after comparisons for the plugins that clearly show these aren't just "snake oil" but rather good sounding basic EQ and Compression (in the case of Viridian 2), at least in my opinion.

Is it better than other plugins? That is entirely up to the end user to decide. Don't get suckered into tribalism or hero woship (listening to "famous names"). Just try the demo and decide yourself.

Also don't get suckered into anti-hero worship and brandhatred just to be part of a different type of tribe. That's even more stupid.

[bmanic]

So, at least in theory, the emulations of Fuse Audio, Cytomic, Overloud and others should be able to emulate the devices a lot more authentically than Acustica Audio ever could.

This very same argument has been done over and over in the physical modeling instruments category as well. Did you ever have the chance to test the original version 1.0 of Pianoteq? Did you ever happen to compare that to a properly multi-sampled piano? Yeah.. that was fun.

Yet some people insisted that Pianoteq 1.0 was way better sounding than any sampled piano as it could do things a sampled library could not easily duplicate (proper pedaling, proper sympathetic resonances etc).. even when the reality of a very weird sounding piano was right in their faces.

Same has been the case for component modeled plugins versus volterra kernels (by the way, not exactly the same thing as dynamic convolution). Once we have every single component properly modeled and every single interaction properly modeled and no constraints on CPU, yes I assume the modeled devices will be way more accurate than any sample/look-up-table based solution. Especially when going "out of bounds" into crazy territory which is impossible to sample with test tones. This _will_ happen eventually and it is already starting to take shape. Companies like Fuse Audio, BlackRoosterAudio, Plugin Alliance (when they employ Ray as the lead coder ), Universal Audio, Softube etc. are already starting to push the boundaries. At the very fringe of "lets do it all properly and research every detail without compromise" lie people like Andrew Simper from Cytomic.. those are the guys that will eventually make all sample based solutions old-school, at which point we'll be purchasing those as "vintage" plugins.

However, even todays version of Pianoteq, in my opinion, still sounds "weird" and not natural at all. There's somethin inherently synthetic about it's sound that just doesn't sound "real". This same kind of synthetic sound has been the curse of older generation analogue emulations as well. Even with all the negatives of Acustica Audio's technology and their plugins (a lot which comes down to themselves not even being able to extract the very best from the technology unlike some of their 3rd party program creators who have a leg up on the mother company) there is a sorth of inherent "realness" that happens to the signal, especially in the plugins that don't do anything crazy.. so mainly their broadband EQs and subtle harmonic enhancements (where the user is not interested in blasting the drive but rather just get some clean harmonics up to the 10th).

So for everyday mixing tasks, gentle operation (as is the norm for most tasks), these plugins work really well. Once you want to truly push things and go experimental (not at all as common as one might think, considering all the hype and articles about magic devices being abused and creating hit records) then this is where the plugins falter and other options become much more viable.

As for the main point of the thread topic: To me it's absolutely mind boggling and absurd that anybody could call Acustica Audio plugins useless or snake oil. That's just preposterous. These are excellent albeit quirky tools. Must have plugins? No, definitely not.. just like anything else. Great to have spices and real time-saver plugins? Yes. Completely unique in plugin land (both sound and weirdness)? Yes, absolutely.

[Ploki]

I don't think the argument was that physically modelled sounds exactly as the original; but that it was as playable. Which I think it's true, there's inherent "organicness" about physically modelled instruments that other approaches lack.

re acoustica: They are unique I'll give you that. I just find it futile and prefer algorithms, but i do so with synths as well, so...

[Digivolt]

Why do Acustica Audio products court so much controversy ?

Both fans and haters are equally quite vicious when it comes to attacking each other whilst those of us who are indifferent are left puzzled why it gets so intense

[funky lime]

Does nobody here print tracks?

If a plugin sounds great but uses tons of CPU, print it and disable. Then you can use unlimited instances.

If you want hardware sound, sometimes you gotta use hardware workflow, i.e. commit to a sound and move on.

Unfreezing is still way less of a hassle than "reamping"