avatar

zmix

Aqua Marine

Posts: 4,058 Member Since:20/01/2011

#23 [url]

Jan 25 17 11:36 PM

weedywet wrote:
I might do another one including q clone just for fun

Yes..! If you have an *actual* 1073 (or whatever the plugins were supposed to be), that would be interesting as a comparo..    

Quote    Reply   
avatar

gtoledo3

Aqua Marine

Posts: 4,110 Member Since:23/10/2013

#24 [url]

Jan 26 17 12:04 AM

This Q Clone looks VERY interesting! I always feel like eq is the weakest link in the box, and that counts for basically all of them, I suppose for the basic reasons you described Chuck.

Can it mimic the static frequency response of a piece of gear that *isn't* an eq?

(I haven't listened to any of these samples in the OP yet.)

Quote    Reply   
avatar

zmix

Aqua Marine

Posts: 4,058 Member Since:20/01/2011

#25 [url]

Jan 26 17 12:14 AM

gtoledo3 wrote:
This Q Clone looks VERY interesting! I always feel like eq is the weakest link in the box, and that counts for basically all of them, I suppose for the basic reasons you described Chuck.

Can it mimic the static frequency response of a piece of gear that *isn't* an eq?

(I haven't listened to any of these samples in the OP yet.)
Yes..!

I've used it to make presets of a few speaker cabinets, and the anti aliasing filter on my Ursa Major Space Station and 224...  it's using LTI convolution, so it's strictly frequency and phase...

I've sent audio into the "Q-capture" return, vocal "ahhs" etc, and you get a convincing formant filter...

To me the really interesting thing is hearing an EQ without any noise or distortion, so my 1982 SSL brown EQ is suddenly a "textbook" EQ, and sounds incredibly good despite the lack of "analog" artifact..    This also means that the D/A and A/D are mostly inconsequential, and that multiple instances, or "adding" layers of EQ within one instance will NOT add distortion, noise, etc..



Quote    Reply   
avatar

demiana

Gold Finger

Posts: 478 Member Since:18/02/2011

#26 [url]

Jan 26 17 3:46 AM

Has anyone tried the Acqua series of convolution plug-ins from Acustica? There is one called Sand which is an SSL channel strip and I'm sure there must be Neves in the range also.

Quote    Reply   
avatar

zmix

Aqua Marine

Posts: 4,058 Member Since:20/01/2011

#27 [url]

Jan 26 17 9:40 AM

Here is an interesting anecdote about convolution based EQ vs algorithmic EQ…

Ages ago I was doing some testing for SSL on their Duende DSP based plug-ins, and at the same time I was doing some work with Focusrite on their liquid mix convolution based plug-ins.

Duende was an algorithmic series of EQ and compressors, and the Focusrite were based on dynamic convolution technology licensed to them by Sintefex.

The Focusrite liquid mix had an SSL  “E” and  SSL “G” series  EQ in  their library, but I noticed that the frequency ranges on their SSL E series EQ  were not the same as the frequency ranges on an actual SSL E series EQ,  but they were identical to the frequency ranges on the Duende’s E series EQ....

Hmmm...

 At this point I thought Focusrite had simply sampled the Duende EQ and not an actual analog SSL EQ  - however - when I  listened to them side-by-side and even do a blind A / B test between the Duende and the Focusrite, with both in their so-called "E series" mode, the Focusrite sounded so much better..!   I could not understand why, because my measurements showed that the curves were identical and there was no distortion or analog modeling being added,  but the Focusrite sounded so obviously much better than the Duende…!

I asked Focusrite if they had in fact sampled the Duende rather than an actual E series console because the frequency ranges did not match an E series EQ but were identical to the E series mode of the Duende.  
They replied initially that they were appalled I would accuse them of misrepresenting their product,  but a few days later they sent a follow-up reply indicating that I was partially correct… they had not used an SSL E series EQ,  - but had, in fact, used the E series position of the SSL 9000 console which has frequency ranges quite different than the SSL E series.

This is when I became fascinated by convolution and convolution based EQ.

It was exactly what it felt like when sampling was introduced to musicians who had previously used synthesizers to emulate instruments, one sounded real and one didn't.

One of the major problems with library based convolution EQs like the Focusrite liquid mix is that the frequency response of each band of each EQ needs to be sampled at every possible cut and boost range in order to give you an interface that allows you to seemingly turn the knobs and get the desired response.  however this method does not allow for the band interactions that sometimes occur in certain circuits notably Trident, Neve and Pultec. The waves Q clone reproduces these behaviors exactly.

Last Edited By: zmix Jan 26 17 9:50 AM. Edited 1 time.

Quote    Reply   
avatar

minister

Gold Finger

Posts: 572 Member Since:27/01/2011

#28 [url]

Jan 26 17 10:07 AM

zmix wrote:


My contention is that since all digital EQ algos rely on the same bi-quad transforms, they all sound identical.  It's not a sound I particularly like, either.
Without the window dressing, the differences are minor, and indeed the only differences available to the designers is the shape of the curve and whether or not to add digital distortion to the algo.

 

I've read this before, and remember a big thread that linked to an article with examples and graphs some years ago. On its face I understand it. But I have never been able to get 2 very different EQ's to sound the same, let alone null. Doesn't mean the statement isn't right.
Chuck, have you tried (and really worked with) Equilibrium or FabFilter EQ?

Last Edited By: minister Jan 26 17 10:26 AM. Edited 1 time.

Quote    Reply   
avatar

Cirrus

Tin Man

Posts: 46 Member Since:08/12/2014

#29 [url]

Jan 26 17 10:44 AM

So, I'm sort of following this. But could someone help me out here; what is an analogue EQ doing that an algorithmic plugin EQ model can't or isn't? Or if it's the other way round of course; what is a modelled digital EQ doing *wrong* that an analogue EQ isn't? I thought it was just frequency response curves and phase shifts, it sounds like we're dismissing any distortions in the analogue circuits as an incidental aside if we're preferring a clean impulse response of an analogue EQ to a modelled EQ algo?

Quote    Reply   
avatar

mario i

Platinum Blonde

Posts: 1,576 Member Since:28/01/2011

#30 [url]

Jan 26 17 12:35 PM

Most analogue EQ will behave slightly different depending on how the circuit is driven from a given source which can vary, and these differences will give it it's character.
There are essentially almost infinite variations in the combination of input source, signal not to mention all the variation in actual EQ setting combinations which
will make up a specific sound.

Whereas most digital EQ's, although modelled, isn't modelled with YOUR input signal, being your snare, guitar, vocal etc...
The way it behaves, is very specific. This is where the CPU power hungry plugs-ins tend to come out ahead of the game.

It's kind of like an 320k MP3 vs 16/44.1 vs 24/48 or 96 vs 1bit/5.6448MHz....

They can all playback music, the question is which is preferable.

M

Last Edited By: mario i Jan 26 17 12:38 PM. Edited 1 time.

Quote    Reply   
avatar

weedywet

Ruby Baby

Posts: 5,864 Member Since:20/01/2011

#31 [url]

Jan 26 17 12:58 PM

zmix wrote:

One of the major problems with library based convolution EQs like the Focusrite liquid mix is that the frequency response of each band of each EQ needs to be sampled at every possible cut and boost range in order to give you an interface that allows you to seemingly turn the knobs and get the desired response.  however this method does not allow for the band interactions that sometimes occur in certain circuits notably Trident, Neve and Pultec. The waves Q clone reproduces these behaviors exactly.

 
yes,
and, correct me if I'm wrong, but that's why if one uses preset 'samples' of say a 10k instantiation of Q Clone, and then another at 3k, and then a third at 100Hz, they don't 'sound like' using all three bands, interacting, on the hardware unit.

 

Quote    Reply   
avatar

weedywet

Ruby Baby

Posts: 5,864 Member Since:20/01/2011

#32 [url]

Jan 26 17 1:01 PM

but, getting beyond the discussion (or tangential to it) of convolution versus algorithmic EQ,

anyone have feelings or prefs on the second set of snippets?

still the same favourites?

Quote    Reply   
avatar

weedywet

Ruby Baby

Posts: 5,864 Member Since:20/01/2011

#34 [url]

Jan 26 17 1:20 PM

G (my 'favourite" eq) is a very different 'flavour' and undoubtedly has different curves.

but interesting that a few people have singled it out.

Quote    Reply   
avatar

Cirrus

Tin Man

Posts: 46 Member Since:08/12/2014

#35 [url]

Jan 26 17 1:41 PM

I've not had a chance to listen to the samples on anything beyond rubbish desktop speakers, but just as a total guess because I know you like it, is G that softube trident eq?

Like I say, a stab in the dark. But maybe if that is what it is I'll win a lollipop.

Quote    Reply   
avatar

Cirrus

Tin Man

Posts: 46 Member Since:08/12/2014

#36 [url]

Jan 26 17 2:11 PM

OK, put on my (still not that great) headphones.

I struggle to meaningfully compare this many samples in parallel - anything more than 3 and I find myself going round in circles, because my impression of one starts being overly informed by whichever two I happened to listen to last,

So, this will seem like stupid methodology but going forward through the clips;

A - I like the top round the snares, doesn't have a lot of depth, softer transient than B. The snare rattle dissapears quite quickly.
B - More brash up top than A, seems more forward, harder transient. Midrange around 2k sounds louder than A and C, yet snare feels smaller
C - More like A than B up top, I prefer the body of C lower down - the low mids are bigger.
D - Closer to B than A and C tonally but transient has less impact, sounds a bit limp
E - Sounds more crunchy.
F - Has less mids, so the treble draws my ear a bit more and the low end seems more controlled
G - snare sits in front of the mix more

Going backward through the clips after flicking between them a bit;

G - Low mids are nice, and the top has nice balance of impact/snare rattle. Sounds big
F - Less snare rattle, less clean to my ears. Sounds smaller
E -Mids around 1-2K draw my ears more - those kinda "A" vowel sounds in the snare body
D - Little smaller than E, more snare than impact.
C - I like how the top end impact and snare rattle sit together, and I like how the body sounds.
B - Harder sounding, more forward but I don't love that. Especially listening to the rest of the kit.
A - top has more snares than impact, sounds more recessed into the mix.

My favourite is C - the way the snare sits in the mix works for me, and the balance of stick impact, snare rattle, shell sound and thump is my favourite.

G would be my runner up, if I wanted the snare to be more forward without necessaraly being louder.

A isn't bad but is just smaller sounding. It's like it's trying to be C but not quite doing as nice a job.
B doesn't sit right to me, the snare is too forward and if I was mixing it I'd probably feel like I hadn't finished with the drum yet.
D doesn't give me the low end I want.
E seems crunchy round the transient.
F sounds a litle more raspy/ reedy to me. No bad thing, just different.

Quote    Reply   
avatar

weedywet

Ruby Baby

Posts: 5,864 Member Since:20/01/2011

#38 [url]

Jan 26 17 7:12 PM

Same settings. I can't swear how accurate their 'setting' is but if I did +2 at 3.2 I did that on all. Except the ones that only had 3k for example. But still set to +2

So it's a leap of faith that when it says 2 it means 2 or when it says 3.2 it means 3.2 etc.

Quote    Reply   
avatar

zmix

Aqua Marine

Posts: 4,058 Member Since:20/01/2011

#40 [url]

Jan 28 17 9:38 AM

weedywet wrote:
zmix wrote:

One of the major problems with library based convolution EQs like the Focusrite liquid mix is that the frequency response of each band of each EQ needs to be sampled at every possible cut and boost range in order to give you an interface that allows you to seemingly turn the knobs and get the desired response.  however this method does not allow for the band interactions that sometimes occur in certain circuits notably Trident, Neve and Pultec. The waves Q clone reproduces these behaviors exactly.

 
yes,
and, correct me if I'm wrong, but that's why if one uses preset 'samples' of say a 10k instantiation of Q Clone, and then another at 3k, and then a third at 100Hz, they don't 'sound like' using all three bands, interacting, on the hardware unit.

 

I shall, not because you're "wrong" per se, but because it's a bit of a straw man argument.

If you use three separate hardware Neve EQs you won't have band interactions either.

If you're using a Q-clone in a real world situation, equalizing your track using your favorite hardware EQ, it will follow the interactions exactly.


Quote    Reply   
Add Reply

Quick Reply

bbcode help