Eiosis AirEQ v1.3

[defjamm]

thx, great info fabrice! i think your product deserves big success, bringing some technical advancement and innovation to the table.
but it could also make one a little angry as a customer. there are so many commercial 'analog-modelled' plug-ins out there, that are far away from their claims and cost big bucks. there are guys using basic dsp algos with nice guis slapped on them, selling stuff like crazy.

@gui: do you have plans to add mouse support to drag into the value-fields?
like with waves ren-eq, there are no knobs, you have your values(frequency, q, gain) and you can directly click on the values and with the holded mouse button you can change them.

imho this works better with plugins and a mouse then knobs. knobs are nice in real-life with some hardware and physical contact, but i don't get a kick out of using virtual knobs.

but if this isn't added no problem, it's usable how it is right now.

[Eiosis]

Nice eq. You know that "not able to save" feature might just be irritating enough to make me buy the thing. But I like that approach to letting us test it out.

In fact we preferred to give an unlimited time of use rather than to stop the plugin processing after a bunch of minutes, to generate a bothering "beep" now and then or to regularly insert a silence. This way, you can test the sound like if you were in a real mixing situation.

Anyway, thanks for the comments !!

i think your product deserves big success, bringing some technical advancement and innovation to the table.
but it could also make one a little angry as a customer. there are so many commercial 'analog-modelled' plug-ins out there, that are far away from their claims and cost big bucks. there are guys using basic dsp algos with nice guis slapped on them, selling stuff like crazy.

I completely agree with you. Contrary to some other companies, we do not exactly say that our EQ is "analog modeled". For me, "analog modeled" means that you have made a model of your electronic board and components, and that this model is so faithful that you must have the same sound. At least, that is what some companies seem to claim.

Firstly, this reasoning is completely foolish and smells too much pseudo-scientifical marketing, secondly, it does not truly works for audio purposes, and finally it is impossible to really achieve without huge amounts of CPU.

What we say for AirEQ is that "AMLT curves are extremely close to the analog reference curves up to 20kHz, whatever the sampling frequency is, with zero-delay latency, like analog does".
This is just the truth.

@gui: do you have plans to add mouse support to drag into the value-fields?
like with waves ren-eq, there are no knobs, you have your values(frequency, q, gain) and you can directly click on the values and with the holded mouse button you can change them.

This will be done, as for keyboard entry of the values. We had no sufficient time to implement all the features that we would do at the beginning of the project, because we spent more time on the algorithms than expected. But we will do our best to implement requirements and wishes coming from the users !


Fabrice,
Eliosound

[eddu]

Does it eat less CPU than URS? I use them all the time because of this

[Eiosis]

Does it eat less CPU than URS? I use them all the time because of this

For the moment, we did not made precise measurements on both OSX and Windows, but we will do very soon. We will compare instances that can be loaded at the same time with several other EQ plugins.

Anyway, the CPU consumption is about the same for URS and AirEQ, maybe a little more for the moment for AirEQ, but some improvements on the CPU will be done for the AMLT Technology in a future version of AirEQ. These improvements will need a lot of work for us, so please be patient !

Many of the AirEQ's users, or many of those who tried it, will tell you that AirEQ is very light on CPU, especially regarding the quality of its algorithms. I beleive that URS uses standard algorithms, that are basically very light on CPU.

Fabrice,
Eliosound

[bmanic]

Hi,

I can confirm that the FL Studio bug is now squashed. Good work! After trying the demo a bit more I start feeling very tempted to buy this thing! Damn u!

Oh well, while waiting for a random sack of moneys hitting my head I'll give you some usability suggestions:

- Make the numeric displays so that they can be dragged up/down for fine controlling the value (this way one doesn't have to sacrifice ones left hand to the evil SHIFT control). Then you end up with two, very close together controls, the knob for coarse tuning and the value for fine tuning. Trust me, it's pretty intuitive.

- Give us an option in the menu for different maximum gain values, like +/- 3dB only or +/- 6dB only. This is of course not a big deal but would again increase the "resolution" of the gain knobs and would be ideal for mastering or buss-eq applications.

Cheers!
bManic

[Eiosis]

I can confirm that the FL Studio bug is now squashed. Good work! After trying the demo a bit more I start feeling very tempted to buy this thing! Damn u!

I'm really sorry..

- Make the numeric displays so that they can be dragged up/down for fine controlling the value (this way one doesn't have to sacrifice ones left hand to the evil SHIFT control).

Great idea ! This will be done very soon. I'll keep you informed about the update. I also don't like the evil, nevertheless, sometimes I use the shift control.

- Give us an option in the menu for different maximum gain values, like +/- 3dB only or +/- 6dB only.

Another great idea !! We will also implement this in AirEQ, and it gives me other ideas. We wanted to make AirEQ with many included options in order to let the user choose the way he wants to work, to display the values, to tune his equalization...
Some features that we wanted to release in the first version of AirEQ are always to be released, because we did not have enough time to work on the algorithms and all the features together, but the AirEQ's workflow will be improved throughout the next versions.

And of course, all updates of AirEQ will be free for registered users !


Fabrice,
Eliosound

[Eiosis]

I was too curious about the behaviour near nyquist and it differs very well from the standard orfanidis implementation.

Yes, it does.
Since our filters are not based on orfanidis, but on AMLT..
I'll post some comparison curve between AirEQ, orfanidis, RBJ and analog.

Here is some curve comparisons between standard digital filters, orfanidis filters, AMLT filters and the analog reference.

Purple : 0dB
Red : Standard digital filter
Blue : Orfanidis filter
Green : AMLT
White : Analog reference


You can see on this first screenshot that even for middle settings, a standard digital filter does not match at all the analog reference, orfanidis is not matched very well and AMLT is matched until 20kHz.



On this second screenshot, settings are more extreme, and it is for this kind of settings that an excellent digital filter curve, i.e. very close to the analog reference curve, is very important for a good sounding equalization.
A warped and unsymmetrical curve makes your equalization sounds metallic and unnatural.

You can see here that it is impossible to match standard or orfanidis filters even for the lowest part of the curve : in this example, the analog reference can be matched until 6-7 kHz, but not further.

AMLT is matched with the analog reference until 20kHz.

Fabrice,
Eliosound

[living sounds]

I'm curious what the analog reference might be...?

Also, since they are matched so closely, what is the difference in regards to harmonic distortion, transient and phase response? If you process a file with the plugin and the analog EQ and do a phase inverted mix of the results, what do you get?

For instance, I created an impulse from the UAD-1 Neve and used this one with a convolution plugin as well as the 1073 plugin on a signal. Although there was a difference (after inverted nulling) of more than 60 db, it was clearly audible in a blind test that the convolution version was inferior. So, when it comes to EQ, a little can matter a lot.

BTW, I think your EQ sounds great (not been able to test it on the monitors, but it performs very well on my laptop headphones), I'm only curious how close it gets to the real thing.

[Barbarossa]

Wow, i'm impressed. I haven't tried this EQ yet, but from the fact, that digital filters differ that much from alanog ones. So the speculation (in the thread "How do you judge an EQ?) can end and we have an important reason, why some analog EQs sound so much better.
Definately have to try AirEQ out. Sadly it uses a dongle.

[warp x]

mmmmmmmmmmmmmmmRTAS

Thanks, just downloaded the demo, will give it a spin later on.

[Eiosis]

I'm curious what the analog reference might be...?

Do you mean "what analog gear did we used to get this reference" ? We could have used sampled IRs of analog EQs to get a reference, but this method has several disadvantages :
- curve warping and phase inaccuracies due to sampling (A/D and D/A conversions, etc..) must be removed to have a perfectly "clean" curve
- you must record many IRs to verify that the reference is correctly matched for all settings
- to get precise settings is very difficult in the analog domain (Gain, Freq, Q-factor)

So we used the theoretical analog reference, given by the transfert function which is used in electronics : H(jw).

Also, since they are matched so closely, what is the difference in regards to harmonic distortion, transient and phase response? If you process a file with the plugin and the analog EQ and do a phase inverted mix of the results, what do you get?

We designed AirEQ to be a "clean musical" EQ, i.e. it introduces no distortion. Distortion introduced by analog EQs may be very musical in certain cases, and very annoying in other cases. Keep in mind that there is also several kinds of non-linearities in an analog EQ, and for example in amplifiers stages supplies, current and voltage limitations may cause some clipping which produces very harsh results in the highs.

Digital equalization does not suffer from power supply limitations, and that's why it can be more convenient to use a digital EQ plugin in some situations. When you switch from a digital EQ to an analog EQ, you have a certain "adaptation time" to understand that you must take care of the analog constraints of the input and ouput levels, and even take care of each band levels. If the level is too low, your S/N ratio will be bad, and if the level is too high, audio signal clipping will occur.
That's the reason why we did not implement a vu meter or a clipping indicator in AirEQ, because it has less utility in the digital domain. Nevertheless, I think that we will implement a level indicator in AirEQ, it may be useful in some situations, for example when you want to chain your EQ with a compressor.

Anyway, we will certainly release an EQ in the future with a less "clean" path, but it requires a lot of research and development to obtain a high-quality algorithm with low CPU power requirements, which, I think, is an important thing for an EQ plugin intended to be used many times in a project.

For instance, I created an impulse from the UAD-1 Neve and used this one with a convolution plugin as well as the 1073 plugin on a signal. Although there was a difference (after inverted nulling) of more than 60 db, it was clearly audible in a blind test that the convolution version was inferior. So, when it comes to EQ, a little can matter a lot.

Total energy of an EQ impulse is very low, so you must take care of the impulse itself and the convolution algorithm that you use, in terms of precision. Depending on the impulse or on the convolution algorithm, you may have some inaccuracies that will result in the differences that you noticed.
We may conclude from this experiment that a little difference in the curve or a little innacuracy in the processing may produce noticeable differences in the sound.

BTW, I think your EQ sounds great (not been able to test it on the monitors, but it performs very well on my laptop headphones), I'm only curious how close it gets to the real thing.

I don't know if we could say that there is an absolute "real thing" . Anyway, feel free to post here your thoughts and comments when you try AirEQ on monitors.

Fabrice,
Eliosound

[KDiamond]

Hi There folks!

Yes, AirEQ is one nice sounding EQ. ALT Max gain mode is a total WIN.

While using Christian's VST plugin Analyzer I came to an interesting conclusion. If you set one EQ to a desired sound and set other eq's to match it's output FFT curve, the sound will be quite close if not identical. At least to my ears when A/B comparing it.

Example.
1. Set EQ#1 to desired sound. Use your favorite one (Algorithmix, PlPar, Hydratone..) - store preset
2. Analyze FFT output
3. Set other EQ's to match EQ1's FFT output. - store preset
4. open DAW, load both EQ's, load presets, A/B compare it

So why buy an EQ if there are freebies that can sound damn close or maybe identical?

The answer I believe is here...

Sound engineer has to get the good sound quick. Tweaking it a long time is anything but good, loosing focus, overall sound picture, objectivity, concentration, energy etc... The satisfying results has to be shown fast! That's why analog gear is so popular. You turn a knob and you're there, thinking it's magic. Same story with guitar amplifiers. If guitar amp has more that 15 knobs it will not make it in history, because guitarist will have too many parameters to tweak. The sound has to be delivered instantly! Well back to EQ's. You must use up to 3 bands to get that "analog" curve using traditional digital EQ's. How's that for a fast work flow? And it's not the same tweaking 1 band or 3 bands in series. It's all about work flow. How fast can sound engineer get to a satisfying results. Longer it takes, worse it gets in terms of his focus.

So it's essential that engineer gets best sound as quick as possible, tweaking as less parameters possible. That's where AirEQ shines. AMLT technology really is damn close to analog curve and that makes it so good. GUI has it's drifts, but as I see Fabrice is very opened for suggestions ant that's very positive. I totally support no visual graphic equalisation curve and was fascinated about ALT Max gain. Genius!!!

Best Regards,
kd

[bduffy]

I'm glad to see this EQ is gaining some ground! This is my EQ of choice now, and I think the CPU is the lowest I've ever seen, especially when you consider the quality of the filters; as low as URS, I'd say, but more forgiving on the high-end. You really can use this on every track - and dammit, I will!

[Eiosis]

Yes, AirEQ is one nice sounding EQ. ALT Max gain mode is a total WIN.

Thanks ! It's a common tip of audio engineers, but maybe developers don't practise enough mixing !
And I believe that we will be copied very soon for this feature !!
More seriously, that is what I call workflow : tools that simplify the life of the mixing engineer, and for a plugin, good ergonomics is what can not be achieved on an analog EQ. If the sound of an analog EQ will remains the reference, tweaking an analog EQ has also some drawbacks, and we tried to avoid these drawbacks in AirEQ by adding many useful features and staying opened to users' whishes and ideas.

Sound engineer has to get the good sound quick. Tweaking it a long time is anything but good, loosing focus, overall sound picture, objectivity, concentration, energy etc... The satisfying results has to be shown fast! [...] So it's essential that engineer gets best sound as quick as possible, tweaking as less parameters possible. That's where AirEQ shines. AMLT technology really is damn close to analog curve and that makes it so good. GUI has it's drifts, but as I see Fabrice is very opened for suggestions ant that's very positive. I totally support no visual graphic equalisation curve and was fascinated about ALT Max gain. Genius!!!

That's exactly the philosophy of AirEQ : to let the sound engineer work fast and intuitively. That's why AirEQ has non-linear controls, has keyboards shortcuts like Max Gain, etc.. I think that an excellent filtering technique or a well-designed workflow alone is not sufficient to get the right equalization simply and quickly.

I will post soon some explanations and examples of my point of view concerning visual vs knob tweaking.


Fabrice,
Eliosound

[KDiamond]

mouse support to drag into the value-fields would be great!

Br,
kd