You are here

Hardware vs Software - Why is There a Clear Difference in this Case?

All about the tools and techniques involved in capturing sound, in the studio or on location.

Hardware vs Software - Why is There a Clear Difference in this Case?

Postby ITHertz » Thu Apr 18, 2019 4:40 am

Hi Folks,

I'm well aware of the endless debates regarding hardware vs software or ITB vs OTB.

A while back I bought a WA273-EQ to use as my main mic pre in my little home studio. I'm very happy with it.

Recently, I came across this article about a simple stereo effect that can be created by setting WA273's HPF to slightly different settings:

https://www.pro-tools-expert.com/produc ... ow-you-how

It's a nice effect but what surprised me more than anything was how poor the plug-in emulation of the same effect sounded (at the bottom of the page).

Is this just an example of something that this particular plug-in doesn't do very well, or is it symptomatic of something more significant? As someone who works mostly ITB but likes analog gear it really surprised me that the difference was so obvious and that the plug-in sounded so bad!

Thoughts?

Cheers,

Chris
ITHertz
Regular
Posts: 101
Joined: Mon Jul 28, 2008 12:00 am

Re: Hardware vs Software - Why is There a Clear Difference in this Case?

Postby Guest » Thu Apr 18, 2019 9:34 am

I think there is a difference if you mix software and outboard. Alex Empire said in an interview that when performing live he put the mix through an old DAT machine because the converters gave a quality to the sound he liked.
When I finish anything I take it to a studio and the engineer puts it through an analogue desk, valve compressor, and analogue eq - and it improves the sound significantly.
Guest

Re: Hardware vs Software - Why is There a Clear Difference in this Case?

Postby DC-Choppah » Fri Apr 19, 2019 2:40 am

When people try to capture the response of a piece of gear, they sometimes use frequency domain response only and forget to include the time domain effects. Then there is the coupled time/frequency domain response, and nonlinear effects, and coupled time/frequency nonlinear effects. There is no end to the complexity of what can happen in the real world and the plugins are only a limited model of that and only include what was modeled.

Lets say that your gear has a flat frequency response. That would be all 1's in the frequency domain response. But let's say that at 10 KHz +/- 250 Hz your gear adds a delay of 10 msecs to the signal. If that delay can be captured in the phase of the frequency response, then it can be modeled in the frequency domain. But at 10 KHz, a phase variation can only introduce delays of .1 msecs. There is no way for a 10 msec delay to be modeled in the frequency domain purely.

So a coupled time domain AND frequency domain response must be modeled. But this is much more complex to do.

Many folks assume that you can capture a real system by its impulse response. But that is naive and only works for systems that have no time domain variations.

Some day we will get past the infatuation with digital and virtual realities and get back to making music in the real world again.
User avatar
DC-Choppah
Frequent Poster
Posts: 1240
Joined: Fri Jul 20, 2012 12:00 am
Location: MD, USA

Re: Hardware vs Software - Why is There a Clear Difference in this Case?

Postby Eddy Deegan » Fri Apr 19, 2019 3:48 am

I've got a lot of hardware (both digital and analogue), and a lot of plugins and I've done a lot of A/B comparisons. I can also play keyboard to a professional standard. I don't judge the sounds on technical metrics, I judge them on what my ears tell me.

Much as the quality of software instruments is really good, I've yet to find a virtual synthesizer that sounds as good as, yet alone better than, the hardware equivalent. The gap has narrowed over the years to the point that you can certainly include softsynths in a mix and nobody will know, but it still sounds better with hardware.

I have no idea why, only a vague hunch that physics are more complicated than logic and software hasn't caught up yet.

One thing I'm not so sure about (and only because I'm not qualified to know) is the difference in virtual processors - compressors, reverb, chorus, EQ etc. I have never owned a hardware reverb unit that can come close to the Aether Reverb plugin for example, but then again I've never owned a truly high-end hardware reverb unit.

All that said, if I was in my early 20s again (nearly 50 now!), and on the budget I was then, I would have loved virtual instruments with a vengeance because they do sound jolly good and are a heck of a lot cheaper.

Every time I see a "you can buy N VSTs with the price you paid for that synth" argument, I shake my head and move on, because there is no convincing some people until they can a) afford the hardware and b) play it competently.

It's not worth losing sleep over, but there is no virtual synth I've ever found that can truly match hardware on a like-for-like basis. It's worth noting though that a good analogue emulation can beat a hardward digital synth doing the same sound in some ways. Digital synths don't respond to player dynamics as well as good analogue simulations do, in my experience. Some come close.

Randomly, I think the filters have a lot to do with it in some cases. In another decade, software might really catch up. My jury is out on that. You'll need a good audio interface though, because software is limited to the capabilities of the host system to physically render those emulated timbres.
User avatar
Eddy Deegan
Frequent Poster (Level2)
Posts: 2836
Joined: Wed Sep 01, 2004 12:00 am
Location: Brighton & Hove, UK
Please consider supporting the SOS Forum Album project, helping charities via music created by forum members.
 

Re: Hardware vs Software - Why is There a Clear Difference in this Case?

Postby desmond » Fri Apr 19, 2019 3:49 pm

There is good and bad hardware and software.

It's fairly easy to demonstrate software which sounds *way* better than hardware, and vice versa, if you cherry pick the instruments. There were plenty of fairly average analog synths back in the day (which are still hilariously overpriced for any musical value they contain, just by being vintage).

Leaving the science aside for a second, we human beings are very much tailored to our senses and our experiences. If you're having a good experience, then you're often in a place where you're enjoying the results more. And sitting in front of a great synth, vintage or not, with a nice dedicated front panel, electricity humming under your fingers and the smell of it tickling your nose (assuming it doesn't just smell of fags and beer, which mostly has the opposite effect), the experience is generally going to be a good one.

I'm 100% sure that if you took, say, two SCI Pro One's, and hollowed one out and put in a dedicated computer running RePro-1, mapped properly to the controls etc, so there was no physical difference between the two units - that no one could pick them reliably apart greater than chance, because you're equalising both the sound *and* the experience. There are a few synths you could do that with - there are great emulations of the Minimoog out there, the Prophet 5, and others.

Sitting in front of a computer, using your eyes to direct your tiny mouse pointer to a part of the screen to move one control just isn't a great experience (although it suits some people just fine), even though the software interface has some advantages over hardware.

If someone lent me a Jupiter 8 now, I'm sure I'd enjoy the experience, and perhaps think in the moment it sounds better than the Roland plugin version. Or maybe not - maybe I'd be underwhelmed at the sound, even though editing it would be a joy because of the tactile interface. But it's not *just* about the *sound*.

The best software synths, and emulations, are totally musically valid and useful, even if the experience is different. And I'm a big believer in going with what best gets you the results you want. It's not a simplistic case of hardware is always better, because it's not. There are great hardware synths, and crap ones, great software synths, and crap ones. It suits people to play to gear's strengths, and use what works for you.

And if anyone does want to lend me a JP8, I'll happily do some tests... :lol:
User avatar
desmond
Jedi Poster
Posts: 8992
Joined: Tue Jan 10, 2006 1:00 am
mu:zines | music magazine archive | difficultAudio

Re: Hardware vs Software - Why is There a Clear Difference in this Case?

Postby CS70 » Fri Apr 19, 2019 3:57 pm

Are you talking of the UAD neve preamp emulation?
It’s not the same as the WAD hardware, so why compare? They would sound different..
User avatar
CS70
Jedi Poster
Posts: 4488
Joined: Mon Nov 26, 2012 1:00 am
Location: Oslo, Norway
Silver Spoon - Check out our latest video  and the FB page

Re: Hardware vs Software - Why is There a Clear Difference in this Case?

Postby ITHertz » Fri Apr 19, 2019 4:11 pm

CS70 wrote:Are you talking of the UAD neve preamp emulation?
It’s not the same as the WAD hardware, so why compare? They would sound different..

I'm only going off the article - they're the ones that did the comparison.

My understanding is that both EQs are only being used for their 50Hz/80Hz HPFs in which case, wouldn't they sound reasonably similar?
ITHertz
Regular
Posts: 101
Joined: Mon Jul 28, 2008 12:00 am

Re: Hardware vs Software - Why is There a Clear Difference in this Case?

Postby Watchmaker » Fri Apr 19, 2019 4:32 pm

The Neve pre amp emulation is awesome imho. but I'd still prefer to have a hardware device.

My theory is that analog is physical and digital is a model of the physical world. It's not surprising then that some models may be less convincing.

I have heard, though do not have deep enough knowledge of algorithms to empirically demonstrate, that the nub of the issue is harmonic content. Pass an electrical signal through any physical component and subtle changes occur to the wave forms, whether that be the carrier wave or the content. This effect is both cumulative and, to an extent, the result of design philosophy. Sample an electrical signal and the ones on zeroes are combined with other ones and zeroes in an attempt to introduce the complexities just described and you get the drift.
User avatar
Watchmaker
Frequent Poster
Posts: 651
Joined: Wed Apr 13, 2016 12:00 am
Location: Upstate NY, USA
Take my advice, I'm not using it.

Re: Hardware vs Software - Why is There a Clear Difference in this Case?

Postby CS70 » Fri Apr 19, 2019 6:20 pm

ITHertz wrote:
CS70 wrote:Are you talking of the UAD neve preamp emulation?
It’s not the same as the WAD hardware, so why compare? They would sound different..

I'm only going off the article - they're the ones that did the comparison.

My understanding is that both EQs are only being used for their 50Hz/80Hz HPFs in which case, wouldn't they sound reasonably similar?

Hm. as I read it, they simply say you can use the same trick (to create a stereo image) with a plugin EQ, not that they are supposed to sound the same. You can probably pull the same trick with any EQ.

I never expect sameness from an emulation, say my La2a sounds great on vocals and so does the UAD plugin, but if I put them side by side they don’t sound the same - and the plugin has four variations! The Warm Audio unit won’t sound like the Neve that inspired it, and neither will another Neve by this time.

But you should get in the same ballpark.. is the plug result so totally different?
User avatar
CS70
Jedi Poster
Posts: 4488
Joined: Mon Nov 26, 2012 1:00 am
Location: Oslo, Norway
Silver Spoon - Check out our latest video  and the FB page

Re: Hardware vs Software - Why is There a Clear Difference in this Case?

Postby Folderol » Fri Apr 19, 2019 6:32 pm

I would say it depends a lot on what you are comparing with what. A modern digital synth is mostly software, so (in theory) should be relatively easy to emulate. A hybrid would be much harder, and a purely analog set up, with no central clock oscillator - no chance!
User avatar
Folderol
Jedi Poster
Posts: 9190
Joined: Sat Nov 15, 2008 1:00 am
Location: The Mudway Towns, UK
Yes. I am that Linux nut.
Onwards and... err... sideways!

Re: Hardware vs Software - Why is There a Clear Difference in this Case?

Postby DC-Choppah » Sat Apr 20, 2019 3:08 am

"One of the main characteristics of Inductor EQ is the way it treats frequencies with respect to time. High frequency signals pass through an Inductor quite quickly however, lower frequencies pass through much slower meaning that audio that is treated with Inductor based EQ exhibits interesting and pleasing kind of phase smearing." from the link in the OP's post.

The plugin version sounds to me like it lacks the time domain response. It was probably modeled with a pure impulse response (frequency domain) only.

The hardware sound is clearly much better. The plugin does not capture this effect.
User avatar
DC-Choppah
Frequent Poster
Posts: 1240
Joined: Fri Jul 20, 2012 12:00 am
Location: MD, USA

Re: Hardware vs Software - Why is There a Clear Difference in this Case?

Postby ITHertz » Sat Apr 20, 2019 6:49 am

DC-Choppah wrote:"One of the main characteristics of Inductor EQ is the way it treats frequencies with respect to time. High frequency signals pass through an Inductor quite quickly however, lower frequencies pass through much slower meaning that audio that is treated with Inductor based EQ exhibits interesting and pleasing kind of phase smearing." from the link in the OP's post.

The plugin version sounds to me like it lacks the time domain response. It was probably modeled with a pure impulse response (frequency domain) only.

The hardware sound is clearly much better. The plugin does not capture this effect.

I think you've hit the nail on the head!

Does anyone else find this disconcerting - with so many people working ITB, that a product from a reputable company which purports to model the circuit behaviour of the hardware actually doesn't?

I don't want to pick a fight with UA but here's what they claim:

Universal Audio’s all-new Neve 1073 plug-in for Apollo and UAD-2 provides all the features, unique circuit behaviors, and coveted sound of Neve's original hardware design. By modeling the dual-stage "Red Knob" preamp, revered three-band EQ, and post-fader output amplifier with obsessive detail, the Neve 1073 Preamp & EQ replicates the experience of the original 1970s hardware with stunning accuracy. Like the hardware, the new Neve 1073 plug-in incorporates all 10 clipping points from the preamp and EQ circuitry, delivering trademark clarity, grit, and harmonically rich class‑A saturation.
ITHertz
Regular
Posts: 101
Joined: Mon Jul 28, 2008 12:00 am

Re: Hardware vs Software - Why is There a Clear Difference in this Case?

Postby Wonks » Sat Apr 20, 2019 10:45 am

DC-Choppah wrote:The plugin version sounds to me like it lacks the time domain response. It was probably modeled with a pure impulse response (frequency domain) only.

I think you meant frequency response modelling only. An impulse response type of EQ plug would capture any frequency dependent phase modification made, as well as the frequency dependent level changes. One reason why a lot of people really like the software products from say Acustica Audio, who use an impulse-based modelling system which does get a lot closer to the original hardware than with frequency-only models.
User avatar
Wonks
Jedi Poster
Posts: 10143
Joined: Thu May 29, 2003 12:00 am
Location: Reading, UK
Correcting mistakes on the internet since 1853

Re: Hardware vs Software - Why is There a Clear Difference in this Case?

Postby DC-Choppah » Sat Apr 20, 2019 4:03 pm

Wonks wrote:I think you meant frequency response modelling only. An impulse response type of EQ plug would capture any frequency dependent phase modification made, as well as the frequency dependent level changes.

But the phase of the frequency response can only delay the signal a little bit. It can't add delays that go beyond one cycle.

If the inductor is delaying certain frequencies by more than this amount it can't be modeled with the phase of the frequency response.

This is a matter of phase delay vs group delay.

It sounds to my ear that they got the 'phase' part right, but not the 'group' part which is much more complex to do right.
User avatar
DC-Choppah
Frequent Poster
Posts: 1240
Joined: Fri Jul 20, 2012 12:00 am
Location: MD, USA

Re: Hardware vs Software - Why is There a Clear Difference in this Case?

Postby Wonks » Sat Apr 20, 2019 4:17 pm

DC-Choppah wrote:But the phase of the frequency response can only delay the signal a little bit. It can't add delays that go beyond one cycle.

I can't see any reason why not? It works for reverbs!
User avatar
Wonks
Jedi Poster
Posts: 10143
Joined: Thu May 29, 2003 12:00 am
Location: Reading, UK
Correcting mistakes on the internet since 1853

Re: Hardware vs Software - Why is There a Clear Difference in this Case?

Postby miN2 » Sat Apr 20, 2019 8:57 pm

DC-Choppah wrote:The plugin version sounds to me like it lacks the time domain response.

I was just about to say something similar as i just did the same thing with Scheps 73 and i the sound is largely similar but there isn't the stereo width of that hardware sample. The software sample in the article sounds pretty different to what i'm getting mind. It's like they forgot to engage the preamp.
miN2
Poster
Posts: 43
Joined: Tue Mar 26, 2019 9:24 am

Re: Hardware vs Software - Why is There a Clear Difference in this Case?

Postby Martin Walker » Sat Apr 20, 2019 9:12 pm

Hmm...

Just had a listen to those audio examples, and am not impressed. Main mistake is that the supposedly mono acoustic guitar recording isn't - I didn't think my ears were deceiving me, and they weren't, since this file already contains stereo information.

Yes, I can hear the improvement in the 2nd file after the 50Hz/80Hz high pass filters on left and right channel respectively. According to the AW manual, these are 18dB/octave, so there shouldn't be that much difference between the two channels, but it's audible and nice. I can also hear the increase in harmonics in the 3rd Warm Audio example, when the channels are driven harder.

However, the virtual 1073 with the 50/80Hz trick sounds completely different - not necessarily a fault, as the 1073 has evolved through a myriad of boutique tweaks over the years. The Warm Audio 273-EQ specifically has Tone switches that change the impedance of the input transformer, specifically to provide tonal changes. From the Warm Audio manual "This switch is labelled 'Tone' for a reason; most listeners will find the unswitched setting to be more open and natural, with engaged setting to be more punchy, aggressive, or thick."

The UAD 1073 could well be modeling slightly different era hardware, but it's stated that "I have recorded an acoustic guitar with a single microphone plugged into a Warm Audio WA273-EQ", and then this recording is passed through the UAD 1073 with the 50/80Hz HPF trick, so it's going through a 1073 preamp twice! :headbang:

I've tried mono-ing the Warm Audio original file and then duplicating the 50/80Hz HPF trick with another virtual EQ that offers separate settings for the left and right channels (rather than passing them through a second 1073-style preamp), and it provided a similar stereo improvement to the Warm Audio original, without the brash tonal change of that UAD virtual preamp in place.

So I really don't think this is a problem for a virtual preamp per se; more some tonal differences between the Warm Audio original and then passing that WA recording through a 2nd Neve 1073-style virtual preamp :beamup:


Martin
User avatar
Martin Walker
Moderator
Posts: 14602
Joined: Wed Jan 13, 2010 9:44 am
Location: Cornwall, UK

Re: Hardware vs Software - Why is There a Clear Difference in this Case?

Postby DaveClark » Mon Apr 29, 2019 1:07 am

Hi all,

Just to be sure we’re all on the same page with respect to the issues being discussed, please consider the following facts about linear, time-invariant systems:

First, consider the real-valued sequence:

1, 0, 0, 0, 0, 0, 0, 0

To construct an IR to delay signals by four samples:

0, 0, 0, 0, 1, 0, 0, 0

one can first construct the complex-valued spectrum (real, imag):

(1,0), (-1, 0), (1,0), (-1, 0), (1,0), (-1, 0), (1,0), (-1, 0)

and take the inverse FFT. A “high-frequency” signal, for example:

1, -1, 1, -1, 1, -1, 1, -1

can be then be delayed by convolution with the result:

0, 0, 0, 0, 1, -1, 1, -1, 1, -1, 1, -1

which is obviously delay of more than one cycle of the high frequency. To delay it by seven samples:

0, 0, 0, 0, 0, 0, 0, 1

one can construct the spectrum:

(1,0), (0.707107, 0.707107), (0, 1), (-1, 0), (-0.707107, -0.707107), (0, -1), (0.707107, -0.707107), (0, 0)

and take the inverse FFT. A delay of any length can be similarly constructed in the frequency domain.

Moral of the tale: You can indeed create delays by working in the frequency domain. It is the coordination of all of the values in the spectra that constructs these delays in the time domain.


As far as RLC circuits go, one can calculate the transfer function in the Laplace domain (Ls, R, 1/(Cs)), then compute an IR that has all delays incorporated. There won’t be any mysteriously missing ingredient that requires time-domain voodoo to make it appear. If something turns out to be missing, it’s because one or more elements or pathways has been inaccurately represented.

Similarly one can construct a solution to the 3-D wave equation in a room starting with the spatial eigenfunctions (that is, with the fundamentals of the room and their harmonics). This technique is demonstrated in books on solving PDEs and in books on Green’s functions. The solution includes the expected initial delay caused by separation of source and receiver, followed by the direct wave (“dry”) spike, then echos, etc.

Contrary to popular mythology, impulse responses do entail time evolution. “Response” means response in time. IRs themselves do not change with time for time-invariant systems, but signals do evolve in time as they travel through a system, as should be obvious.

Again, I'm just trying to ensure that we are all clear about the properties of LTI systems.

Regards,
Dave Clark
DaveClark
Posts: 1
Joined: Mon Apr 29, 2019 12:55 am

Re: Hardware vs Software - Why is There a Clear Difference in this Case?

Postby resistorman » Mon Apr 29, 2019 5:09 am

Plugins are great. I’ve sold most of my “color” preamps and record everything as clean as I can. With modern converters I don’t have to worry about compression going in, just record with plenty of headroom. Any modern cpu will run many times more plugins than hardware processors I had back in the day. Even if they were only 80% as “good” who cares? But they are much better than that. You can’t even start an argument about processors that were digital in the first place. Instead of worrying about how many angels can fit on the head of a pin, we should be capturing the best performance possible and mix it with whatever is at hand, which is likely far better than what was available in those “vintage” days of yore...
User avatar
resistorman
Frequent Poster
Posts: 678
Joined: Sun Nov 22, 2015 1:00 am
Location: Asheville NC

Re: Hardware vs Software - Why is There a Clear Difference in this Case?

Postby Guest » Mon Apr 29, 2019 10:10 am

Eddy Deegan wrote:Much as the quality of software instruments is really good, I've yet to find a virtual synthesizer that sounds as good as, yet alone better than, the hardware equivalent. The gap has narrowed over the years to the point that you can certainly include softsynths in a mix and nobody will know, but it still sounds better with hardware.

I basically agree with you and have never got on with soft synths - but there is something I don't understand. When I was learning to to programme Serum, a software wavetable synth, I sampled a sawtooth wave from a Clavia Nord 3. It sounded better, and punchier, in Serum than on the original hardware synth. When I need to use a synth sound next I will sample more waveforms and see if that is usually the case or just the exception. Serum is the only software synth I can get on with and feel comfortable programming.
Guest


Who is online

Users browsing this forum: No registered users