What Makes Hardware Reverb Better?

[rackdude]

So, I hear all of this talk about hardware reverbs, and am wondering how, since it's digital, could sound better than software verbs since they are both digital. I mean, is it that you are running it through analog stages? So how about running software reverbs through a preamp or something?

Also, I was recently thinking about pulling out my old Alesis Multimix 8 and just running a feed into it for the Alesis reverb since a lot of people seem to have their reverb hardware, do you think that's worth doing if I already have plugs? I don't want to go through the trouble if I don't have to

[rackdude]

No comments?

[biggator6]

fwiw - I've always wondered the same thing... dedicated processor in an outboard box vs. more powerful processor in the computer - technically there's zero reason it can't be made to sound just as good (unless the reverb is using some custom-spun DSP chip.. then something like the UAD/duende/liquidmix/powercore should work)

[Jorg]

its all just down to the algorithm.
If Lexicon used their algorithms in a plugin then surely it would sound just as good as the hardware. However they wouldnt sell anymore hardware units tho.

Question is:

How powerful are the DSPs in e.g a PCM 96 or Bricasti? Could a modern studio PC easily handle those algos or would you need a DSP card?

[Dean Roddey]

It's not just an issue of whether the PC can handle the same algorithm, but whether it can do that while doing everything else it's having to do. If it was dedicated to just reverb, I'm sure it could. But if it's doing everything else as well, maybe the DSPs in the hardware unit do have an edge, I dunno. The folks creating a hardware DSP can use every CPU cycle available, while the folks creating software reverbs have to compete on reasonable CPU usages as well as sound.

I'm not saying that means hardware is inevitably better, just pointing out some of the issues.

[Jorg]

Originally Posted by Dean Roddey

It's not just an issue of whether the PC can handle the same algorithm, but whether it can do that while doing everything else it's having to do.

Thats obviously what I meant.
Would be interesting to know how many instances of a vsti with a Lexicon algo a 8 core Mac could run. That could answer the question if you actually need hardware to process High End Reverbs or if Hardware is nothing more than a big Dongle.

[Casey]

Originally Posted by Jorg

How powerful are the DSPs in e.g a PCM 96 or Bricasti?

The DSP capacity of the Bricasti M7 = 5 Digi Accel Cards.



-Casey

[Jack Ruston]

Well I think there are two main factors here.

1. The sheer power available in eg a bricasti, is far in excess of what an average desktop can be expected to devote to reverb. Some of the newest systems are very powerful but they have a lot of other things going on.

2. The greatest expertise in the manufacture of digital reverb algorithms lies with the people who have been doing it the longest. I.e lexicon, bricasti, tc, eventide etc.

[rackdude]

Well, how much dsp is there in the old lexicons for example? It's hard for me to think a box from even the 90's could have as much cpu as a Pentium 4. CPUs doubled power so quickly!

[Dean Roddey]

1. The sheer power available in eg a bricasti, is far in excess of what an average desktop can be expected to devote to reverb. Some of the newest systems are very powerful but they have a lot of other things going on.

That's probably true as a standalone fact. But the real question would be, how much CPU do you actually need to create a very good reverb. If the answer is less than a Bricasti has, then it may not matter. Though, even if that were true, it could still be more than what a general purpose system can dedicate to the process.

However, you also have to consider how long that would be true if it was true. It won't be that long. With the huge advances being made in the data paths and caches of multi-core/die systems and the ongoing increases in the individual core performance, I'm not sure it'll be an issue that much longer, even if it is now.

Yeh, you could always make a dedicated system that's more powerful, but it would only sell if that extra power was worth the much larger cost. This is the same issue that will face Digi as well. Yeh, they could always make a more powerful system, but the costs will go up excrementally, and people won't pay a large amount of money for power they don't actually use.

[robertshaw]

Originally Posted by rackdude

So, I hear all of this talk about hardware reverbs, and am wondering how, since it's digital, could sound better than software verbs since they are both digital. I mean, is it that you are running it through analog stages? So how about running software reverbs through a preamp or something?

Also, I was recently thinking about pulling out my old Alesis Multimix 8 and just running a feed into it for the Alesis reverb since a lot of people seem to have their reverb hardware, do you think that's worth doing if I already have plugs? I don't want to go through the trouble if I don't have to

The algorithms are virtually the same when comparing highend hardware to highend software verbs. The difference is the hardware verbs have fatter sounding analog i/o. Since SW verbs don't have any analog
components they tend sound thin. Less thin if you have great DA though I'm sure.

But I think Waves IR1 or Alitverb with the right impulse is as good if not better than hardware. Some of the high end lex verbs and even the h3000 verb is great but those devices were always used in an analog domain
with an analog console and analog tape. No shit they sound better.

I bet if you used an analog console and used an aux send into your pc for just verb it would sound fatter than just ITB. as long as you have a good a/d d/s. If your using cheap stuff like berhinger or rme sure it's all gonna sound thin shit in shit out

All harwdare verbs have an analog portion as well as a a/d /d/a portion. Not to mention they have a processor. Essentially a PCM 90 or whatever is a computer it just does on thing and only runs one
software app so to speak

[Casey]

Originally Posted by Dean Roddey

With the huge advances being made in the data paths and caches of multi-core/die systems and the ongoing increases in the individual core performance, I'm not sure it'll be an issue that much longer, even if it is now.

Speaking just from the point of view of Bricasti algorithms; It's very difficult to get away from dedicated memory systems. I don't see this changing anytime soon.

These algs just do not play nice with others that wish to share the same memory. Whether it be L1, L2, L3 or main memory. I remember expanding on this in other threads.



-Casey

[thethrillfactor]

This questions comes up once a month now. If you really research it out you'll discover that it really comes down to 3 things.

The 3 things are:

1) The algo's

2) The harware and processing power that needs to be dedicated

3) The designer's ear who makes the patches that will eventually use the above and make the processor legendary.

1) The Algo's-

If you research and study the history of effect processing alot of the algos used today are borrowed from years past when some of the greatest & most creative minds were in there heyday. The reason i say greatest and most creative is because alot of the early pioneering work of digital effect processing came out of fields that had nothing to do with audio like hospital & heart monitoring equipment(Lexicon) and guidance systems for the Avionic industry(Eventide). So it would take special and unique minds to apply the limited technology to try to do an incredible and complex task which is to recreate the complexities of rooms & spaces. Out of these times came men like Dr. Barry Blesser(who developed the algos for the first digital reverb the EMT 250), Anthony Agnello(who engineered the first digital pitch shift processor & the algos for the now famous SP2016), the nuclear physicist/classical engineer Dr. David Griesinger(who was the first to merge a microcomputer with his reverb algos and came up with the Lex 224), Christopher Moore(who was responsible of bringing cost effective digital reverb to the studio in the Ursa Major Space Station by taking different taps at different points on a delay line and feeding it back into itself there by not needing alot of onboard RAM which back in 1978 was really expensive).

Basically these were designers that even though were limited by the technology at that time were inspired to do what had not been done before and make it work in the studio musically at the same time.

Do we have these kinds of minds working at the software companies currently making the reverb algos you are using in your plugs? Probably not. Right now these kinds of minds are applying their mental efforts in other fields. Also lets not forget the EMT 250 when it was first developed put together the great technical(Dynatron & EMT) scientific minds(Barry Blesser/Karl Otto Bader) across two continents to bring it to fruition. And if you were lucky to own one you paid heavily for it-$20,000 back in 1978!!!! Its no wonder it still holds up today as probably the best sounding digital reverb processor.


2) Hardware and Processing-

If you check out the link below by Casey of Bricasti Design he best explains it what the power is needed to get a plug in to do what a hardware processor can do:


http://www.gearslutz.com/board/new-p...-fx-pcm96.html

By the way in the Bricasti design you can hear that these guys really get it.

And lets not forget in the older reverb designs that have become our favorites, the limitations have become part of the sonics. These things can't be modeled in software and specs wise would be looked down on by most modern designers.


3) The patches;

Lets face it...if the patches on any processor, be it hardware or software suck and are not useful it will not sell. Period. It takes someone with a creative mind and a really good ear to do sound design for the processors. I remember someone posted a question here why does the SPX 990 sound different than all of the other SPX series and the answer is obvious. What Yamaha did was instead of letting their team of scientists who had previously made the patches for their previous products do them for the 990, decided to branch out the development to their team in England to get a different spin and perspective. What the english team did was developed the ideas and got input from different recording engineers all over Europe. From the feedback they fined tuned it and voila the 990 has been a staple in live and studio rigs for years.

Some people just have the ear for sound design and others don't. Don't know why that is though, but this fact can make or break a processor.

[Chaellus]

What Makes Hardware Reverb Better?


they sit in the mix better

[Aisle 6]

Originally Posted by robertshaw

The algorithms are virtually the same when comparing highend hardware to highend software verbs. The difference is the hardware verbs have fatter sounding analog i/o. Since SW verbs don't have any analog
components they tend sound thin. Less thin if you have great DA though I'm sure.

But I think Waves IR1 or Alitverb with the right impulse is as good if not better than hardware. Some of the high end lex verbs and even the h3000 verb is great but those devices were always used in an analog domain
with an analog console and analog tape. No shit they sound better.

I bet if you used an analog console and used an aux send into your pc for just verb it would sound fatter than just ITB. as long as you have a good a/d d/s. If your using cheap stuff like berhinger or rme sure it's all gonna sound thin shit in shit out

All harwdare verbs have an analog portion as well as a a/d /d/a portion. Not to mention they have a processor. Essentially a PCM 90 or whatever is a computer it just does on thing and only runs one
software app so to speak

I do not think that this is correct. For instance, I run my high end reverbs through a completely digital path. They are patched through a Z-Sys digital detangler and then digitally inserted into PTHD. They still kick arse over most plug in verbs. I am no programing expert, but I can only assume that it has much more to do with the code itself and the DSP. Having said that, there are some nice plug in verbs, and they all have their place. It is just for a more detailed main verb, I seem to always end up back at the hardware.

[Mr. PC]

Excrementally?

Tell us how you really feel!


PC

[Dean Roddey]

If you look at where were are today, where a quad core system, at least for those doing the traditional thing of recording incoming audio as apposed to running lots of very large soft synths, can carry a massive load, and you project out a few years to an 8 core system, where each of the cores is considerably more powerful and efficient than the ones we have now and all the other data flows are faster, the caches are larger, and so forth, we are going to be heading into a era where plugs can be come considerably more expansive in terms of the CPU they use, because it's no longer an issue having just enough to get by.

If you can do what you need to do now with a quad core, then an eight core is basically another four cores of gravy, that can be used for better modelling, more powerful and realistic reverb algos and so forth. An eight core system where each of the cores is 1.5x as powerful as the ones in the quad is an even further jump.

And project out, say, five years out when you are on a 16 core system, each of which is, say, three times more powerful than the one you have now, and think of what that's going to mean. You could afford to dedicate cores (and their L1 caches) to particular tasks if the DAW allowed for that.

A 16 core system where every core was only twice as powerful as the cores you have in a quad now would be 8 times the power you have now. Think about what kind of power that implies. There's some loss as you add more cores of course, but five years out the cores will probably be more than twice as powerful as now as well, so the real increase will probably still be more than 8x.

[narcoman]

Originally Posted by Casey

The DSP capacity of the Bricasti M7 = 5 Digi Accel Cards.



-Casey

cool!!! nice to know some people still do have big dicks thumbsup

[zboy2854]

Obviously Bricasti is among the best of the new breed. But one has to think that a still legendary verb such as the Lexicon 480L, the processing power of which is now nearly 20 years old can easily be replicated in software verb if the issue is simply processing power or CPU cycles, given how much more powerful today's computer DSP is.

So why isn't there a software verb that is universally agreed sounds as good as the 480L at least?

It can't be the DSP horsepower, we've long since eclipsed the horsepower of a 480L in the box.

Which narrows it down to the algos or the converters.

Given that many folks have run 480L's or other hardware verbs digitally into and out of their DAW's, bypassing the converters, and suffer no ill effects, that eliminates the converters as the reason.

Which leaves the algos.

[thethrillfactor]

Originally Posted by zboy2854

Given that many folks have run 480L's or other hardware verbs digitally into and out of their DAW's, bypassing the converters, and suffer no ill effects, that eliminates the converters as the reason.

Which leaves the algos.

The 480L, Roland R-880, Quantec, Dynatron 255(EMT 250) Eventide 2016 & H3000, and Ursa Space Station all sound different when run digitally.

They don't sound as lush or as wide or blend as well in the mix. In all of these cases the reverbs sound better when converted to analog. Alot of these boxes were designed in an age where everything was interfaced with a console, not a computer. Also alot of the converters were specially designed for each box. It was all one system, not just a box designed for its own.

[rackdude]

I wonder what it would sound like if someone q-cloned the analog path of the reverb and then used a software reverb with the q-clone infront.

[robertshaw]

Originally Posted by Aisle 6

I do not think that this is correct. For instance, I run my high end reverbs through a completely digital path. They are patched through a Z-Sys digital detangler and then digitally inserted into PTHD. They still kick arse over most plug in verbs. I am no programing expert, but I can only assume that it has much more to do with the code itself and the DSP. Having said that, there are some nice plug in verbs, and they all have their place. It is just for a more detailed main verb, I seem to always end up back at the hardware.

I used to own a 480L a 224 pcm 60 and 70. Waves ir1 sounds just
as good just no registers to mess with. I don't now what plugins you use but I would suggest looking into alti asn waves ir1. also get the lexi impules. Maybe try updating your converters mainly D/A the reverb will sound as good as hardware. But if hardware woks for you then keep using it. whatever makes you happy.

But the algos for hardware verb/sw verbs have not changed in 35 years. The only difference is hardware (including pc hardware) can now process these 35 year antique algos in real time.

All a 480l was, was and a/d converter -> a computer that processed the input with the reverb Algorithm then converted back to analog What do you think a computer does that has an a/d d/a converter a DAW and reverb plugin. It's doing the same thing. Everyone knows lexicon units had incredible conversion and analog components for their era.

The only thing I miss about my old lexi units was the paramaters and register tweakability. But with the new impulses I feel the reverb sounds much more realistic. There are so many impulses now just pick one and make a record

Checkout waves IR1 with the lexicon and Eventide impulses youll be more than satisfied. Also if you are using stock PTHD D/A that may not be helping the sound of your plugins at mixdown. Upgrade to something phatter like a hedd unit or a Burl and everything will sound bigger

[quantumpsych]

all i need to know is that Nigel Godrich uses convolution verb. if the algos have not changed in 35 years, it's just a matter of converters and analog processing. the convolution technique was probably not even imagined 35 years ago because the dsp did not exist?

[Aisle 6]

Originally Posted by robertshaw

I used to own a 480L a 224 pcm 60 and 70. Waves ir1 sounds just
as good just no registers to mess with. I don't now what plugins you use but I would suggest looking into alti asn waves ir1. also get the lexi impules. Maybe try updating your converters mainly D/A the reverb will sound as good as hardware. But if hardware woks for you then keep using it. whatever makes you happy.

But the algos for hardware verb/sw verbs have not changed in 35 years. The only difference is hardware (including pc hardware) can now process these 35 year antique algos in real time.

All a 480l was, was and a/d converter -> a computer that processed the input with the reverb Algorithm then converted back to analog What do you think a computer does that has an a/d d/a converter a DAW and reverb plugin. It's doing the same thing. Everyone knows lexicon units had incredible conversion and analog components for their era.

The only thing I miss about my old lexi units was the paramaters and register tweakability. But with the new impulses I feel the reverb sounds much more realistic. There are so many impulses now just pick one and make a record

Checkout waves IR1 with the lexicon and Eventide impulses youll be more than satisfied. Also if you are using stock PTHD D/A that may not be helping the sound of your plugins at mixdown. Upgrade to something phatter like a hedd unit or a Burl and everything will sound bigger

I have been using the Waves IR verb for a few years and like some of the impulses. I am using the Lynx Aurora 16 converters. I will give the converters a go again and have a listen. Thanks for the input.

[aflon-telom]

Originally Posted by Casey

Haha, Next time you go to the theatre, if you order popcorn, you will notice that it comes in small, medium, large, and my dick.



-Casey

[switches off imagination right now]

[living sounds]

IRs sound muddy, static and ugly in the high end compared to my hardware Lexicon units (connected digitally as well). The latter works far better in the mix as well.

[markus enochson]

i think real rev boxes are hard to beat with plugs still...

[Washington]

Originally Posted by Casey
Haha, Next time you go to the theatre, if you order popcorn, you will notice that it comes in small, medium, large, and my dick.

OMG... If their products are half as cheap as their humor, it's the bargain of this century. Still laughing though, sure takes balls to be that lowkey here. Can't wait to lay my mitts on it (the M7, I mean)

[jamwerks]

I remember reading somewhere that the new convolution reverb "Mir" from Vienna Symphonic library, in it's developement stages, was putting a multicore PC in the red with just one instance.

Goes to show that maybe computers will never be too powerfull.

[jamwerks]

Originally Posted by Washington

Can't wait to lay my mitts on it (the M7, I mean)

LOL