Convolution software; the situation so far...

[Michiel Post]

There is a growing choice of convolution software packages (some are integrated in DAW software), some eat up most of your CPU resources, some have processing delays of>300 ms. Who has experience with several of them and what are your preferences so far?

[Simone]

There is a growing choice of convolution software packages (some are integrated in DAW software), some eat up most of your CPU resources, some have processing delays of>300 ms. Who has experience with several of them and what are your preferences so far?

Talking about the released ones: definitely Waves IR-1, the sampled room sets are marvellous and the quality and flexibility is stunning...

Just my 2 euro cents ;)

Ciao,

Simone Coen

[Lazul]

definitely Waves IR1 for tweakability and IR library.

[P de Caumette]

definitely Waves IR1 for tweakability and IR library.


Hard to beat Voxengo Pristine Spaces (http://www.Voxengo.com) as far as quality/price is concerned.
You can get excellent I.R. at www.noisevault.com to complement it.

[Simone]

Hard to beat Voxengo Pristine Spaces as far as quality/price is concerned.
You can get excellent I.R. at www.noisevault.com to complement it.

I kinda agree, but the library soming with IR-1 is not even comparable... real theathres and gourgeous spaces recorded at 32 bit float / 96 KHz resolution with the future chance of true surround reverb (any surround format, any!)

Also: tweakability is unbeatable... like an 'old' digital reverb...

Best,

Simone Coen

[Freya]

www.catt.se - I really love the sound of this program. I've been working with a simulation of Mechanic's Hall in Worcester Ma for a while now and it really brings the sampled pianos to life. I'm just about to start playing around with its 5.1 upmix capabilities.

Freya

[T Parks]

I've got to agree with Simone. The IR1 is one of the best bits of software I've ever had the pleasure of using. What sold it to me was the amount of editing that can be achieved compared to the other products. There's so much tweakability that, if used subtley, can make a big diference to the mix. The amount of decent halls and rooms etc is quite staggering, I've not got halfway through them all yet!.

[PeterRoos]

I have Pristine Space and think it is absolutely great, especially for its low price.

But I have to agree that there is a huge lack of really good natural spaces, which is where IR-1 jumps in, with its high price. For pro's IR1 is probably the best choice, but for amateurs and semi-pro's PS is a very good alternative.

I have a set of impulses from Ernest, which are the best I have heard, but they are unfortunately not yet in true stereo-stereo (quad channel).

The free impulses on NoiseVault are nice, at best, but are not (yet) up to par with the impulses from Ernest and probably Waves (I have only used the evaluation version with the embedded impulses).

I think it is time for AudioEase to release a commercial library of true stereo impulses for use in other convolvers than AltiVerb. I would gladly shell out some 200 dollars for a good set of natural spaces.

[Bruce A. Richardson]

I agree with everyone here so far, that Waves IR-1 is absolutely the best I have used. The library is amazing, and the flexibility and tweakability of the impulses in real time is best of class.

[Sovereign]

I agree with everyone here so far, that Waves IR-1 is absolutely the best I have used. The library is amazing, and the flexibility and tweakability of the impulses in real time is best of class.Any tips on creating a convincing soundstage with it, Bruce? Which impulses would be the best to use? I haven't had much time yet to delve deeply into IR-1.

[fade up]

I've started using IR for the first time (Space Designer) and I had my eyes opened - wow. Too new to make any comment on it, just posting to say this section of the forums will be read by me with interest.
Thanks.

[Bruce A. Richardson]

Any tips on creating a convincing soundstage with it, Bruce? Which impulses would be the best to use? I haven't had much time yet to delve deeply into IR-1.
They're all really great. You can hardly go wrong with anything in the package.

[JohnCarter]

i don't think IR-1 is so good... there is a lack of real BIG reverb
The recording stage impulse TAOD is good for Brass but not big...
But i like the lexicon reverb impulse (Dark Hall, Gold Hall )

[Simone]

Hi JohnCarter...

well... if you're looking for a looooong reverb I guess yuo should look elsewhere...

first: convolution reverbs eat for CPU cycles the longer the reverb tail is
second: those extra long reverb belong to DSP emulations more than to real life situations... loooong reverb are usally found in badly reverberant places (like car parks, garages...)

IR-1 features some of the world's best renowned spaces... that's it...

Just my 2 euro cents...

Best,

Simone Coen

[JohnCarter]

ever heard of EWQLSO sample library ? is it possible to have a hall reverb like EWQLSO with IR-1 ? ( to avoid some answers : I KNOW that we can't reach that reverb quality with artificial reverb ! )

[Simone]

of course I know about it...

The halls 'sampled' in the IR-1 and other convolution reverb do not have anbything to envy to the EWQLSO hall... IMHO...

the EWQLSO hall is very midrangey and colored IMO and it is ok for orchestral stuff buty not ok for a lot of other styles IMHO... in IR-1 halls I find a lot of more useful things... again IMHO...

Ciao ;)

Simone

[JohnCarter]

thanks for your answer

and excuse me , just a last question Mr (Like Lieutenant Columbo, you know ? )
you seem to have a lot of knowledge in impulse, reverb etc... so what is the secret of the hall of EWQLSO ? i'm trying actually to simulate (not to reach his quality level though) i through IR-1 with the toad recording stage impulse but i'm far from it actually. If you want you can listen to the percussion ewqlso platinum demo (which is available in 3 file, close mics, surround mics and full mics) there

EW-155

(at the bottom "Percussion" (all 3 mic positions) - (T. Bergersen). To download it )

[Simone]

well... I have the EWQLSO so I know it quite good (I wrote a review about it also...)...

The secret now is: the EWQLSO qas recorded in a real hall, while applying IR-1 to recording as it is now will allow you for the placement in a limited way since you'll be using the same 'sample' position for all of the instruments... soon convolution reverbs (and for sure IR-1, since I know personally the guy who did the convolution stuff, Professor Farina) will allow for placement of instruments in the virtual convoluted space... I think that will be the point where you will be looking for dry recorded samples so that you can realistically immerge them in a real acoustic environment 99%... and personally I'd rather leave 1% reality in change of the chance of changing reverb type, lenght and so on...

Ciao,

Simone Coen

[Houston Haynes]

I think that will be the point where you will be looking for dry recorded samples so that you can realistically immerge them in a real acoustic environment 99%... and personally I'd rather leave 1% reality in change of the chance of changing reverb type, lenght and so on...

Ciao,

Simone Coen

I've heard some vendors ascribe to the same theory, and I'm saving up for a dual Opteron system in order to do all of that in real time. ;) But in the mean time, I'm also willing to deal with the trade-off - in fact, there's a way to get that next 0.5% closer with currently prevalent computer systems, but will keep it to myself for now... :D

[Bruce A. Richardson]

This forum has managed somehow to propagate a totally incorrect and false assumption that reverb units cannot supply the lush, long tails and fanstastic spaces heard in commercial recordings.

Well, that is bunk.

The lush, long tails and fantastic spaces heard in commercial recordings are produced ALMOST EXCLUSIVELY with digital reverb.

They're also produced by expert, lifelong, professional engineers at the very top of their game.

You don't just get a computer, buy some software, and become an engineer. Even some really great musicians are poor engineers at best. This is why I always tell people to forget mixing while they're composing--then forget the composing process while mixing. These processes NEED to be separated, because the mindset needed to do the work is a completely different one.

Almost every commercial recording has four to five professional "hands" in it before you ever hear it. The tracking engineers are experts in microphones and in recording the ensemble at hand in a way that can be constructed into a plausible "whole," by the mix engineer. The artists are most involved at this stage, making sure their "sound" is being captured realistically and pleasingly. The producer is also weighing in at this stage. Next, a mix engineer generally takes over the process--whose expertise is entirely centered in the tools and techniques of mixing projects. At times, the mix engineer is going to call for some re-tracking, if there are tracks that are just not going to work--and those are subsequently re-cut. The producer is heavily involved by this point, and weighing in on ther overall "slant" of the sound. Finally, when this is done, the work goes to yet another set of ears, the mastering engineer, and again, some of the work may end up back in the mix engineer's studio if the mastering engineer cannot make it work as a final, cohesive, project.

NOW...

Contrast that with the DAW user experience. In general, most DAW users have never even been in a real studio, much less done work in one. So the sound source from the very outset is likely not recorded optimally, nor recorded in an optimal room, nor recorded with an optimal mic plot and signal path--OR, comes from samples and samplers where all those decisions have been made in advance...and which cannot be INDIVIDUALLY TAILORED to the project at hand. That is extremely important in and of itself, because in the commercial recording process, "getting the sound" of each instrument being tracked is THE NUMBER ONE CONTRIBUTOR TO THE END RESULT. For instance, the way the piano sounds in a mix is more determined by where the mics are placed than any other factor after that. And the mic placement is arrived at by having the artist playing the track while the engineers and producer listen against the other tracks and work with placement after placement until the TRACKING SOUND is almost ideal--before one iota of mix work has happened. In fact, before the PERSON who mixes the album has arrived many times.

The second issue with most project/DAW situations is just a plain lack of mixing and engineering experience, and out of that grows a misuse of some of the tools as often as not. EQ gets misused most often by trying to boost elements not being heard rather than to remove what's preventing an element from being heard. Compression is so misused that people actually refer to it as a bad thing--yet it is the most useful (and widely used) mix tool in the box for both decreasing and increasing dynamic range as needed to achieve good mixes.

And reverb. Where does one even start? If the imaging has not been tweaked to near perfection by means of tracking techniques and basic mixing techniques before reverb, then reverb will never, ever, ever, EVER make that happen.

Forget audio for a moment, and think about film. Even if exactly the same components make up a frame of film, a different lens dramatically affects the way that picture looks, how it focuses, how one element relates to another, and how the whole of the frame will relate to the next setup you see. Story is being told via the lens itself.

AND...once you've shot your frame, really the only control you have over is is to mush it up. You either filter it to make it softer, to highlight one aspect at the expense of another, to make it skew somehow...but every action you can take from that point on is mushing something.

Likewise, if you're a fan of painters, the painters we consistently reward with "master" status through history are not mushers...they're "bold strokers."

Sorry, this is going to make a point sooner or later.

Where I think people are thinking wrongly is that they're depending upon reverb to accomplish this long litany of tasks in the total imaging of audio which has nothing to do with reverb. Imaging is 90% set by the time you've recorded the track, and you are going to do very little to affect it--in a way that you'd call "real," with any reverb.

In the case of the Waves IR-1, I guarantee you that if you are hearing a dissatisfying result with that tool, then you are trying to bend a track WAY too far from its original image to possibly sound good. Because IR-1 is quite excellent--probably the very best convolver you can buy at this time in terms of both flexibility and the outstanding library.

Yes, we have advanced tools now. But they don't change physics and acoustics, nor do they supply any shortcut to mastering the engineering processes that are required to do world class mixes. I sure don't consider myself a world class engineer by any stretch of the imagination, but I have worked with enough of them to know that blaming the tools is never the answer. A great engineer can do more with levels, EQ, and compression than mere mortals can do with all the gear in the world (which essentially comes with any DAW you buy).

[Simone]

I totally agree with Bruce!

best,

Simone Coen

[Michiel Post]

I really like this controvery: some sample libraries are recorded with up to 3 recording perspectives (Close, player, ambience for piano or close, ambient + hall for orchestra). These recording perpectives are Fixed Microphone Positions based on good assumptions and (often) years of recording experience.

One thing that struck me right from the start is how consistent the EWQLSO library sounds without any additional reverb beeing used. The recorded ambience and hall is well chosen and works very well. If you go to their demo section you will see that almost everybody used this library without any additional reverb. Why?!

It gets complex when you try to mix the pre-recorded perspectives with "other" pre-recorded perspectives libraries. Then you record a couple of tracks with several such libraries and when you get into the mixing process, opening a pandorra's box of rerverberation systems, you soon find that the pre-recorded perpectives are fine for each instrument in solo but don't combine well with the other sample libraries choices. So you get a new problem; which libraries combine well?

I didn't buy the WAVES IR verb but did a few tests with the demo version and I found it almost as good as the quality Altiverb stuff. Both seem to have a library that is well worth the purchase and offer the flexibility to resolve complex tasks as the above one.

I would love to have some of the IR's of the EWQLSO hall to see how well these could help setting up a convincing solution of certain problems. I even think these IR's could be worth a lot of $$ when they do what I expect them to do.

On the other side of the spectrum you have the close or even anechoic recorded instruments. Once they feed a signal in an IR-based convolution verb you have a whole new experience; you can actually place the microphones yourself (after the recording of the samples) which gives you other options to balance ensembles.

I like these new technological improvements.

[sunra]

"The lush, long tails and fantastic spaces heard in commercial recordings are produced ALMOST EXCLUSIVELY with digital reverb.

They're also produced by expert, lifelong, professional engineers at the very top of their game.

You don't just get a computer, buy some software, and become an engineer. Even some really great musicians are poor engineers at best."

My sentiments exactly!

Rick :cool:

[JohnCarter]

"They're also produced by expert, lifelong, professional engineers at the very top of their game."


I agree with that

One big example that prove this affirmation is a demo made with Vienna Symphonic Library :

it's a mockup of a Hollywood Soundtrack called "Air Force One", and it's incredibly well mixed ! You know who mixed it ? "Mal Luker" may be you don't know this name, but it's a professionnal engineer who worked on severals hollywood soundtracks, the result :

http://vsl.co.at/admin/downloader.asp?file=/data/Sounds/MP3/Air%20Force%201.mp3

http://vsl.co.at/admin/downloader.asp?file=/data/Sounds/WAV/Air%20Force%201.WAV

[JohnGrant]

Verb freak that I am, I got to know: what's the difference (aside from features) between the IR 1 and SIR?

JG

[mike harper]

interesting thread, i was wondering what you guys think of sam's room simulator? how does it compare to the waves ir?

[Per Lichtman]

Verb freak that I am, I got to know: what's the difference (aside from features) between the IR 1 and SIR?

JG

There is a fair amount of discussion of that very topic at the following links. Hope it helps.

http://www.noisevault.com/elephanttalk/viewtopic.php?t=419&highlight=ir1

http://www.noisevault.com/elephanttalk/viewtopic.php?t=480

http://www.noisevault.com/elephanttalk/viewtopic.php?t=392&highlight=ir1

[JohnGrant]

Thanks for the info; I'm going there.jg

[christian marin]

john, glad you found out the *not so hidden* link ;-)
sorry for having to remove it - we are no longer entitled to publish it.
best, christian

ps: bruce, great - your parable to film and painting

[Nick Batzdorf]

I sure don't consider myself a world class engineer by any stretch of the imagination, but I have worked with enough of them to know that blaming the tools is never the answer. A great engineer can do more with levels, EQ, and compression than mere mortals can do with all the gear in the world (which essentially comes with any DAW you buy)

Yup. That's one of the things that impressed me right away about a friend of mine who definitely is a world-class engineer: he just uses what he has without complaining, and it always sounds good even if it's a Digidesign 001 and a cheap condenser mic.

The only thing I might question is the part about tracking and mixing engineers being two different people. I mean, yeah, they often are, but I think every good mix engineer - with the exception of mixers who come from a re-mixing background - knows how to record as well.

The two are different skills, sure, and in many ways mixing is the more creative of the two, but it's hard to imagine very many really good engineers who aren't able to do both.

Having said that, if "you" (generically, as in "Joe/Josephine Musician with DAW") will be shocked at the difference if you ever get to mix something and then hear what a really good mixer does with it. Your mix will probably sound good, but his/hers will sound better.

[Nick Batzdorf]

(not dissing remixers, of course)

:)

[charles]

Yup. That's one of the things that impressed me right away about a friend of mine who definitely is a world-class engineer: he just uses what he has without complaining, and it always sounds good even if it's a Digidesign 001 and a cheap condenser mic.


I sure don't consider myself a world class engineer by any stretch of the imagination, but I have worked with enough of them to know that blaming the tools is never the answer. A great engineer can do more with levels, EQ, and compression than mere mortals can do with all the gear in the world (which essentially comes with any DAW you buy)

this is exactly the same situation with samples and the art of orchestration,
it seems to me that it all comes down to the inner ear, if you can conceptualize the orchestra accurately in the inner ear your orchestrations are bound to work when they reach the outer world, i.e. real life situations, that is why a well orchestrated piece using GM sounds will sound better than a badly orchestrated VSL piece, i know nothing about compressors etc, but i would imagine the inner ear has exactly the same relationship in handling eq's compressor's etc. it all comes down to the appropriate intrinsc geniuness of the original ideas in the inner ear

[Jake Johnson]

I have an older copy of Samplitude (Studio) that has hall spaces that can be applied. I never seem to see this program mentioned when people write about the subject, although I think this program was one of the first to include the feature.

Are they using an older technology, or are their samples not as good as more recently recorded ones?

[panzerD]

I have an older copy of Samplitude (Studio) that has hall spaces that can be applied. I never seem to see this program mentioned when people write about the subject, although I think this program was one of the first to include the feature.

Are they using an older technology, or are their samples not as good as more recently recorded ones?

Good question. I would guess that it's partly due to most people here using other sequencers and want a plugin. But I am curious how some of the people here think Samplitude's impulses compares to the others.

I have heard some great pieces at the VSL site that use Samplitude.

[chriscaouette]

What about Prosoniq Rayverb? I am on a mac and use EWQLSO Silver and was looking at this one or even possibly using 8 channels of light pipe to send to a pc running Pristine Spaces.
Any thoughts?
Chris

[Jake Johnson]

(Correction: I have Samplitude Project, a reduced version that came with my EWS 88 card, but still has convolation.)

Actually it has both Convolution and Room Simulator. The Convultion dialog box is minimal--it requests that you load a file and then apply the convolution. Apparently the effect is better this way than if you use the many templates\room samples in the Room Simulator.

Some confusion here for me, though: In the new pianos, is the convolving applied to each note, or to the entire sound generated by the multisample?
(The entire sound, since otherwise each note would need its own sample of the room space\inside of the piano, with mics placed at different places to register the verb as it sounds at different places in the piano?)

[Michiel Post]

The sustain pedal convolution is only applied to the samples played while the sustain pedal is DOWN. All other notes (with pedal up and release triggered) are free of convolution.

[Jake Johnson]

One souce of confusion for me, here:

My impression is that convolving is intended to be a substitute for sustain. But convolving, if I understand it, is a mixing of the sample of a soundspace's verb with the sample of a note. If this is correct (and I understand that I may not understand), would convolution work with pedal up samples, too, since they are also being played in the same sound soundspace?

[Layne]

One souce of confusion for me, here:

My impression is that convolving is intended to be a substitute for sustain. But convolving, if I understand it, is a mixing of the sample of a soundspace's verb with the sample of a note. If this is correct (and I understand that I may not understand), would convolution work with pedal up samples, too, since they are also being played in the same sound soundspace?

Someone correct me if I'm wrong, but my assumption is that the pedal-down impulse would include the sympathetic vibrations from the other strings(?)

I would think if you used convolution on pedal-up it would be a different impulse.

[Jake Johnson]

Yes, that's correct--it would need to be a different impulse. I think: At least I'm assuming that the soundspace reverb is recorded with the sus pedal down, but that may not be right: with the pedal down, overtones would sound, differing depending upon what was used to create the verb--a pistol shot would make certain strings ring in sympathy (for whoever the bullet hit?) and other sounds would create other vibrations. So maybe the soundspace sample is done without the pedal down?

Would the ideal thing be to record a separate sound space for each actual note and each velocity (since the hardness of the strike would create different numbers of vibrations in the wood and affect the speed at which the vibrations reached the wood and then bounced back and forth inside the piano) and use the actual hitting of the note to create the source of the soundspace. Would take many hours of recording though.

[hv]

The 64K question in my mind is, "How many IR's are needed?" The more I think about it, the more seriously I doubt that a single IR will be anywhere near as realistic modeling the entire pedal-down range compared to individual pedal-down samples. I could be totally wrong but my intuition tells me that if you fired a pistol shot directly over the hammer above middle C, you'd have to pick up different overtones and reberberation than you would at either one of the far ends of the scale. Different nearby strings. And the nearby body walls are shaped radically differently. But, available memory often being the constraining factor, there'd be a tradeoff to consider between the possibility of having more pedal-up velocities and a less accurate pedal-down sound.

I suspect that 88 IRs sourced above each string might be indistinguishable from sampled pedal-downs. If that held true across velocity layers, a single set of impulses might replace multiple layers of pedal-down samples with a pretty substantial memory savings. But this being pretty new technology, we'll probably just have to try out the possible alternatives and see.

Howard

[Jake Johnson]

(bump)

I'm surprised no one has responded to these concerns. Has anyone experimented with differing mic placements for each sample of groups of samples?

[tahome]

What about Prosoniq Rayverb? I am on a mac and use EWQLSO Silver and was looking at this one or even possibly using 8 channels of light pipe to send to a pc running Pristine Spaces.
Any thoughts?
Chris
Rayverb is very cool, but it's not exactly an "IR playback reverb". It constructs a "virtual room model" from the IR which you can then tweak (extensively!). I use it with pre-recorded IRs mainly to have access to existing spaces and then change their size, width, wall material etc. The cool thing is that you can use the automation to do all this dynamically, which is great for post-processing foley or dialogue tracks.

--th

[Guga Bernardo]

Someone correct me if I'm wrong, but my assumption is that the pedal-down impulse would include the sympathetic vibrations from the other strings(?)

I would think if you used convolution on pedal-up it would be a different impulse.

No, that´s not correct. When you sample a pedal up note the sympathetic vibrations are also being sampled! In fact you can hear them very clearly in most libraries that have pedal samples, so there is no real point in adding them again (!) as a convolution.
Aplying a convolution reverb over pedal down notes may sound good, but it has nothing to do with the real acoustic behavior of the piano. Modeling body resonance is a completely different thing than convolution, because it requires the emulation of every single aspect of the acoustic of the instrument, and not only the behavior of the sound in a given room.

[Nick Batzdorf]

The catch is that you have to upgrade to HD, of course. :-)

[Houston Haynes]

One thing that might have been mentioned in Bruce's post and those that followed about having "many chefs" - but I missed while skimming the thread - is that sometimes a project is made better simply because a fresh set of professional ears have come into the room. So, it's not *just* that the skills are tuned to a specific job, but that the ears walk in with a fresh perspective. I'm a conservatory-trained composer with a degree in audio engineering, and I *still* get someone with good chops and ears to check my work at the mix and master stage - and when I can, I hire out the last stage altogether.

To me, the specificity of skill means that you can hone into the correct sound and result in a shorter period of time, which prevents steam from coming out of the ears of the producer. ;) Aside from being lighter on the checkbook, limiting the time-on-task by a specific function also limits the number of opportunities for a tired set of ears to drive things into the ditch.