Ravel,
For what it's worth, i've performed in the SOH in both the Concert Hall and the Opera Room. The placement profile provided by the computer generated rings that suspend from the ceiling create an extremely tight 'space' almost per instrument.
I sang Peter Allen's 'I still call Australia Home' (to my eternal shame) in fornt of the SSO, with no mike. I was stage front and approx 5 metres behind me was the front row of the Orchestra going back approx 15 metres to the W/W. (It's a big stage).
I could hear myself sing (I don't know if that's good or not) comfortably, and the audience had no problems either. And the astonishing part is, the balance obtained between instruments is almost perfect, with the singer or soloist hearing each part as clearly as the audience, with no monitor required.
It is a unique building, and a real master class in computer aided acoustic design. The sound is quite dry on stage, indeed singers often need a drink handy, but the sound for the audience in all parts of the Concert Hall is full without that middling 'porridge' so apparent in older more conventional designs.

I would suggest a more 'regular' resonating hall for an example, as the SOH is truly one of a kind, and doesn't perform 'reverbally' as one would expect.

Thinking about it, you could almost say the SOH Concert Hall is a 'hardware' version of our expectations of MIR!

Regards,

Alex.

Dietz: In general, though, wouldn't one assume that an impulse is intended to be used in-line unless specified otherwise?

Also, convolution processors tend to have other controls for adjusting the wet/dry balance, for example Altiverb lets you adjust the level of the tail and ER.

Am I wrong?

Of course you are right that you can adjust ER and reverb tail in Altiverb (and TLSpace, and IR1, and others). This is just a question of clever implementation of time- and envelope-dependant changes to the IR.

But don't forget that there is a third part in most IRs - the direct part _before_ everything else happens. For example: I didn't use Waves IR1 much due to some strange installation-problems, but AFAIk this engine simply cuts the first transient when you import an impulse-file. This "direct" signal carries the signature of the whole recording chain, nonetheless, that is why you better get rid of it an replace it with the original dry signal, most of the time. This means: You are better off with a "traditional" Auxiliary-Send/Return-scheme, even using sampled reverb.

Especially when you sum a high number of IRs for a single source things can get ugly due to minmal (sub-sample) delays and phase-issues. - I think I won't uncover a trade secret now when I tell you that the MIR's concept will take exactly these issues into account.

Of course there _are_ scenarios where you absolutely _want_ the colouring of the signal chain, like IRs of EQs and the like. But this is not what I'm after when I'm using sampled reverb.

HTH,

Wow Dietz,
That was deep.
I think something was lost in the translation.
But thanks for giving your 2 cents.
RAVEL

Hm - you're right, English is only my third language and not my mothertongue. I fixed a typo, but this obviously wasn't your problem in understanding my message ...?

It was perfectly coherent to me!

That's interesting about the first samples being cut off so you can use a send-return. What i don't quite follow is why subsequent samples don't also carry exactly the same information about the recording chain. The transient response wouldn't have any practical effect, but surely the same recording chain captures the whole impulse response? In other words the frequency response is going to be affected throughout the whole sine wave sweep, isn't it?

What's not penetrating the fortress?

And do you happen to know what the the Audio Ease guys are doing with their ER positioning feature? It works really well - I'm sure not as well as MIR, but it's very effective.

I don't know too many details about AudioEase's own stuff. As I said before, we didn't simply "take AltiVerb to PC", they build a dedicated engine for us and our specific needs.

Regarding the signal chain of an IR: Of course you are right that the whole recording carries the signature of the equipment and techniques involved. But as our incoming dry signal is already "recorded", I would like to avoid yet another colouration (unless I'm looking for a certain, additional effect). But this isn't even the main reason to get rid of the direct part inherent in each IR (because we are able to get rid of this signature by brute-force math): If you combine a _lot_ of different IRs for a single source and combine the dry part of each and every of them, too, you will most likely get some slight phasing problems; they wouldn't expose themselves by combining the "wet" parts only.

HTH,

Oh, of course. Yes, you already have the ERs on the samples.

I wasn't thinking.

ok - trying to keep up with you guys. right now i use altiverb and mostly the sydney opera house irs through aux/return 100% wet. since the dry part is already present in the main signal, under what circumstances would it ever make sense to move away from 100% wet in altiverb? isn't balance "best" acheived by making wet/dry adjustments at the board?

@Martin Bayless said:

ok - trying to keep up with you guys. right now i use altiverb and mostly the sydney opera house irs through aux/return 100% wet. since the dry part is already present in the main signal, under what circumstances would it ever make sense to move away from 100% wet in altiverb? isn't balance "best" acheived by making wet/dry adjustments at the board?

Yes, of course, if you go the typical Aux-Send-route. But I think we were discussing the idea dropping the dry signal alltogether from the final mix.

@Nick Batzdorf said:

Oh, of course. Yes, you already have the ERs on the samples.

I wasn't thinking.

I'm not talking about the early reflections - they are crucial for the signature of the room and the instrument's behaviour in it, as well as the positioning cues.

If you analyze a typical IR you will most likely see a short amount of silence (reflecting the distance from the impulse to the microphone), then a razor-sharp spike (the dry signal, which can be extremely low in volume, of course), follwed by another short amount of silence. Only after this initial phase of the IR the actual reflection-part begins, most likely with the ERs from the floor at the very beginning, only fractions of a second after the remaining dry spike. About 100 to 300 ms later you already have the dense reverb tail, which carries only little positioning or disctance cues any more and sounds pretty similar from many source-positions in good music-facilities.

So you wouldn't want to lose the ERs, but only the first few milliseconds. Another way to achieve this is to keep the intrinsic timing (IOW no cutting at all), but to silence the remaining dry part within the IR-audiofile.

All these aspects will be considered and integrated in the MIR-engine when it comes to maximise flexibilty on the one hand, and performance on the other.

Anyone try the "Media Ventures Live Room" setting in Gigapulse?

I was wondering, since this setting does not really have reverb, could I use that to simulate the recording of an orchestra in that room, and then add another reverb on top?

Yes, I could imagine that a typical way to use this would be to "put all instruments in a room together", and add some synthetical reverb an the top of it, like a rich, modulating Lexicon L960, or one of the crystal-clear tc System 6000-algorithms.