SIR means "Super Impulse Reverb". :-] - a real-time convolution reverb for PCs.

It's a shareware(?) VST-plugin programmed by Christian Knufinke, at the moment available as a free beta-version at [URL=http://www.knufinke.de/sir/index_en.html]this site[/URL].

I cannot point out often enough that finally, the quality of a convolution reverb is mostly dependant from the impulse responses used - and at the moment there's nothing better than Ernest Cholakis' IR-sets, available from www.numericalsound.com .

HTH,

/Dietz

Hello,
Just to report back that I've tried SIR (with Ernest Cholakis' impulses of course - yes these are fantastic [:D] ) and in the end I bought Samplitude instead to run the 'room simulator' in there (also a convolution reverb). Ernest Chokakis also recommended Samplitude. It offers some very useful control, and for some reason (which he couldn't explain either) sounds 'better'. (There's a free demo at Samplitude.com).
But SIR is free of course... [:)]

All the best,

Simon

... and of course, there are several other options for convolution-reverb, but most of them are not realtime, like Acoustic Stamp from Steinberg's Nuendo/Cubase SX, or the Acoustic Mirror from Sonic Foundry's (now Sony's) SoundForge.

/Dietz - Vienna Symphonic Library

Dietz sure is right about the quality of impulses - I am trying to use Acoustic Mirror but the impulses that Sonic Foundry has are very disappointing. I'm going to try the Cholakis impulses because so far nothing has sounded as good as the Lexicon hardware reverb I already have.

It depends - for artificial, "pop-y" (if this is a word) reverb tails, there is nothing that surpasses machines like the tc System 6000 or a Lexicon L960, or when you need control over a lot of parameters, like in post-pro.

For true rooms which stand for their own, convolution reverb is miles ahead of any "modelled" reverb. Try Cholakis' IRs (or our MIR, if you're patient enough) and you'll hear what I'm talking about.

/Dietz - Vienna Symphonic Library

Dietz could you give a few pointers on any differences in how you treat a convolution reverb differently from an old-fashioned reverb? For example in creating a realistic concert hall ambiance. Do you use more wet signal and try to select the convolution that sounds at 100% like what you want? In other words if it needs to be drier, you select a drier sounding hall model rather than turn up the dry signal on a wet sounding hall. I get the impression that is one main difference between the two.

Good question, William.

The main difference is that we have much fewer options to change the sonic signature of a convolution-based reverb than with the "classic", synthetic reverbs we are used to - unless someone takes the step to manipulate the IR _itself_. This can lead to some major brain surgery :-] if you know what to do, or you make things simply worse in case you don't. All the Sampling Reverbs I know, no matter if they are hard- or software-based, offer just a few basic parameters to tweak, like IR-length, pre-delay, some more or less sophisticated ways to change the envelope of the IR, basic EQ and/or filters, and maybe stereo-width.

With synthetic reverb, you have a myriad of options, down to the position, the level and the timing of a single early reflection (out of 1024 or so ... 8-] ...). This is great, and a big problem at the same time, because you have to know pretty well what you're after - and how to achieve it.

This means: yes, it's mainly a question of choosing the right set of IRs, and a more or less intuitive adaption of the few aspects of their sounds I mentioned above. While this leads to a behaviour I call "preset zapping" (something I actually hate, wearing my sound-engineer's hat), it is actually a very easy and conclusive way to find the room I like (... from the perspective of the musical half of my ego).

The mix between dry and wet levels is _very_ specific and depends on completely un-scientific aspects like personal taste, the mood and the arrangement of the piece I'm working on, as well as on the source and the quality of the IRs used, of course. But in general I'd say that I use more of the wet signal in my mixes when I use Sampled Reverb - simply because there's no reason to hide its deficiencies (as it is the case with synthetic reverb, most of the time).

****

I like this thread! :-]

/Dietz - Vienna Symphonic Library

I'm currently using Ernest-verb [:)] in Acoustic Mirror as follows:

Dry with no gain reduction.
Wet at -9 to -12 for most of the orchestra. I'll use -3 to -6 for anything I want to be really ambient.

Depending on how busy the piece is, I'll generally use an envelope set anywhere from 2 secs to 0.8 seconds. This allows me to use higher amounts of wet signal without the result getting muddy. For very ambient sounds I'll disable the envelope.

I also have about a 50% reduction in the stereo width of the dry signal and occasionally rolloff the high-end.

This is what works for me on the project that I've been on for the last few months - which generally has moderate to brisk tempos.

Dietz,
Thanks a lot for those ideas. I see what you're saying about the adjustability vs. artificiality. I have never wanted to get into adjusting all those parameters and like the whole idea of convolution because maybe you could say it bypasses them by using the charactacteristics of the sampled place instead (?)

Slaroussels -
that info is great because I just ordered the Cholakis impulses and am going to attempt using it on Acoustic Mirror.

Could you clarify these things - "dry with no gain reduction" - do you mean it is at maximum? Or just at 0? I believe there is a possible plus 20 db on the plugin window I have.

Your envelope setting - is this a change from default in acoustic mirror? Do you think that is necessary? I like to avoid changing defaults unless absolutely necessary, but see what you're saying about avoiding mud. That has been one of my main problems since I'm trying to get a very richly ambiant hall sound.

I have done no reduction in stereo width except on an instrument specific basis - for example, collapsing the horn section in image size simply because it is not that big in the orchestral context. Again, do you think this is necessary and what does it accomplish? I 've been disturbed by reverbs that did not seem to reverberate throughout the whole space, but instead in the area where the instrument was playing.

Do you roll off the high end of the dry or wet? I would be reluctant to roll off the wet because it would alter the room characteristics, but maybe its necessary in practice.

Sorry to ask so many things but as you can probably tell I am obsessed with this right now!

Thanks,
William

Could you clarify these things - "dry with no gain reduction" - do you mean it is at maximum? Or just at 0? I believe there is a possible plus 20 db on the plugin window I have.

Just at 0.

Your envelope setting - is this a change from default in acoustic mirror? Do you think that is necessary? I like to avoid changing defaults unless absolutely necessary, but see what you're saying about avoiding mud. That has been one of my main problems since I'm trying to get a very richly ambiant hall sound.

I have no idea how Mr. Cholakis did these impulses, but bear in mind that most impulses are captured in an empty hall. This give a much longer RT60 (i.e. reverb length) than a hall filled with people, so I think it's not unrealistic to use a shorter verb length. Also, if you listen to most classical recordings, you aren't going to notice reverb tails of 8 seconds which is about average for a Numerical Sounds impulse without using the envelope feature. As always, just do what sounds good to you.

I have done no reduction in stereo width except on an instrument specific basis - for example, collapsing the horn section in image size simply because it is not that big in the orchestral context. Again, do you think this is necessary and what does it accomplish? I 've been disturbed by reverbs that did not seem to reverberate throughout the whole space, but instead in the area where the instrument was playing.

I actually collapse these on a per instrument level with the DSP Station mixer within gigastudio. This is necessary if you want the ensemble to sound further back in the mix, which seems to be more consistent with the sound of the Numerical Sounds impulses - way back. I'm really looking forward to the "stage mic" type impulses in MIR.

Do you roll off the high end of the dry or wet? I would be reluctant to roll off the wet because it would alter the room characteristics, but maybe its necessary in practice.

I roll it off the dry - that seems to be the best way to feed the Numerical Sounds impulses.

Slaroussels

Thanks a lot for that info - that makes perfect sense about the envelope and roll off. It actually sounds like something I can handle, not being an engineer. My previous recordings have been done with a Microverb, Zoom, or Lexicon reverb and turned out sounding too artificial to do justice to the VSL samples (or even previous libraries) no matter how I tried to adjust things like dry/wet, EQ, etc. So I'm hoping that by getting a basically natural sound to begin with there won't be so much need of adjusting and tweaking every little parameter.

William

Just for the record -

AFAIK, the lenght-values supplied by Ernest Cholakis / Numerical Sound are not to be mistaken as the RT60 we are used to interpret as "the reverb length", but actually the raw sample-length of his IRs. As RT60 they would look much shorter.

/Dietz - Vienna Symphonic Library

Dietz,

Could you explain why that is? Is it because the audible reverb is shorter than the measureable? Also, would you agree with these settings we're talking about on these parameters? Of course much of it is subjective I know, but as a basic approach...

Thanks,
William

are the Ernest Cholakis IR's useable on altiverb?

Anybody know?

David T

Ask Ernest, you can email him by clicking on the Contact link on his website http://www.numericalsound.com

@Another User said:

AFAIK, the lenght-values supplied by Ernest Cholakis / Numerical Sound are not to be mistaken as the RT60 we are used to interpret as "the reverb length", but actually the raw sample-length of his IRs. As RT60 they would look much shorter.

True. Though they are still plenty long. Which is not a bad thing - it's easy to use the envelope to make it shorter. Making it longer, on the other hand...

I think above all the point is that really good quality reverb contains a lot of very low-level detail. The RT60 refers to a standardised 'cut-off' point beyond which we say the reverb tail is 'finished' - but on Ernest's impulses he's carefully preserved the barely-audible information beyond that, hence very long files. This is good news; also it means it's definitely worth getting the 24 bit impulses, and worth using 32-bit floating point audio where you can, to make full use of this low-level information in the convolution calculation.
I'm still getting to know the various 'classical spaces' on the CD - each one seems to have its own very individual character.

It's good we have this forum - I'd maybe never have found out about any of this otherwise!

All the best,

SImon