Mir Demos - Great ... but quite dry?

[Trailerman]

Hi

I've been listening to the Mir demos  for the first time today.   I'm exceptionally impressed with the depth of sound, clarity and realism on both of them.  By far the most realisitc reverb I have hard applied to VSL arrangements, ever.

However, to my ear they both sound quite dry, and I wondered if I could ask Dietz (who I gather mixed them) whether this was intentional, whether it's a result of certain settings choices (like listening position), or whether the rooms which have been modelled and are represented in these demos, just have relatively short reverb times.  In my experience, the average concert hall (and I realize that's already a contradiction in terms) tends to give the listener a longer RT than you hear (for example) after the final organ notes have stopped at the end of Zarathustra.

Is it perhaps because the tracks are mixed from the conductor's position, rather than a listener further from the stage, and does Mir offer the ability to move the listening position as easily as the instrument position?  Even more revealing, would be some alternate mixes, where the listening position was further from the stage, just to get an alternative take on what Mir can do.

Many thanks for any feedback

Jules

http://www.trailermen.com

[Dietz]

Thanks for the positive feedback, Jules. Some very interesting topics you raise here.

The demos posted on our site are just that - demos. They were made according to the very personal taste of all people involved (Christian Kardeis as the composer/arranger/programmer, Martin Saleteg and Florian Walter as MIR-developers, Herb Tucmandl as The Final Vote ;-), and myself). We all agreed that we didn't want to overdo it and tried to find a convincing balance between the direct signals and the room information MIR adds to them. The funny thing is that I heard a few complaints that the pieces are too wet already, so it really boils down to individual preferences, it seems :-) ... It would have been an easy task to shift this balance, on behalf of the general Dry/Wet-Offset feature, but the mixes your hear are simply what we came up with :-) ... no additional sweetening applied!

The listening position of "Zarathustra" is the one from the 7th row, not the conductor's point-of view. Changing that is as easy as selecting another set of IRs (i.e.: another mic position) from the Select Venue-dialogue window. You wouldn't even have to re-arrange your virtual orchestra on the stage. - Another option is to move the main microphone virtually, although this feature is meant to fine-tune the perspective of your listening position in relation to the instruments on stage, and does not alter the room-information per se.

At the moment, the reverb-times you have within MIR are exactly as they appear naturally. Due to the fact that we deal with several hundreds, sometimes more than thousand of IRs for each and every stage, we have found it to be impracticable to alter the length of a room like you would expect it from simple convolution reverbs. This might change with a future version of MIR, but right now, you would either select a different hall if you aim for longer reverb (just as in Real Life ;-) ...), or add some overall reverb to the "natural" mix, like it is done more often than not with actual orchestral recordings, too.

Doing alternative mixes is a good idea, actually. Maybe I can arrange something, but please don't hold your breath for it.

... I hope this answers at least some of your questions.

[Trailerman]

 Dietz - I'm extremely grateful for your detailed and illuminating response.  You've answered all of my questions and more so!

You're absolutely right of course, that the amount of reverb is very much a matter of personal taste, and it's clear from your account that you didn't simply plonk the instruments in a virtual space and leave it at that, you tweaked the sound and settings to deliver what you felt was the right balance of clarity and ambience - as anybody would do.   Clearly the capabilities of Mir allow plenty of refinement of a given 'space' and it's interaction with the component parts of an arrangement, and if I understand you correctly, it would not be hard to supplement the 'Row 7' IR's with an alternative set from a different listening position.   I'm assuming therefore that each modelled space included, offers a number of different IR sets which one can use depending on taste (might you be able to reveal how many sets per room - roughly?).

I've always felt being able to tamper with IR length in convolution reverbs was something of a contradiction - you present a space as accurately as technology allows and then enable that presentation to be tampered with - so I'm not concerned about the ability to alter reverb length, it strikes me as a purist approach.  I am interested in the ability to virtually alter the mic position though.   If you move the mic further away from the instruments, how does Mir compensate for the additional distance, without being able to change the nature of the IR set?

Anyway, I think it sound sgreat, and I'm already looking into an additional system as a 'Mir Machine', in anticipation of its release.  Just so I'm clear, would I then also need to run all my VSL instruments (VE instances) on that same system, or is Mir designed to run 'standalone' on it's own rig, with VE being used to pipe audio around via LAN?

Thanks again for all the additional info Dietz.  This is a very exciting time for virtual orchestration.

Regards

Jules

[Dietz]

Thanks again for your positive feedback, and sorry for the delayed answer to your questions, Jules -

Your assumption is right: In its first version, Vienna MIR is supposed to be used as a (more or less) stand-alone "black box" with integrated VI-player. The VIs opened "within" MIR receive MIDI-data, either from a second computer or a sequencer running in the background on the same machine; the latter setup would mean that you sacrifice some performance power for this task, of course - power which would maybe better be used for MIR itself.

Regarding the possibilty of moving the microphone virtually: This feature is made possible due to the fact that _all_ signals within MIR are kept in Ambisonic format - a "meta-format" which allows for free choice of the number of virtual microphones, their directivity patterns, the angles between them and so on. Together with some cunningly implemented "distance scaling" of the direct signals' volumes, it is possible to give the listener the impression that the microphone "moves" in relation to the stage. But this movement is happening _within_ a selected set of impulse responses. This means that you will perceive (e.g.) a broader stereo imaging with increased amount of direct signals as soon as you move the mic closer to the stage, or a re-adjustment of the stereo balance when you push it towards the left or the right side, but without actually changing the IRs themselves (other than decoding them differently from Ambisonics to your chosen output format, that is).

HTH,

[Austin]

Hi Dietz,

So in your opinon if one were to use MIR and a sequencer on the same machine, would a single quad Nehalem 2.93 Ghz with 12 GB of DDR 3 1600 be adequate for running an orchestra? Would a dual Xeon Nehalem 2.66 with 12 GB of DDR 3 1600 MHz make a much bigger difference?

[Dietz]

Like stated before: We strongly suggest the use of processors of the Intel i7-family for the use with Vienna MIR.

[wolfgang]

@Dietz said:

Like stated before: We strongly suggest the use of processors of the Intel i7-family for the use with Vienna MIR.

Dietz, I believe the Nehalem processors Austin is mentioning are, in fact, server-grade xeon i7 processors (released just last week) I think.  I assume these would be as good or perhaps even better for MIR than the consumer i7s, yes?

[Austin]

Hi Wolfgang, you are correct in your assumption. The Nehalem processors that are Xeon are server grade i7's but they are referred as Xeon 5500 and not i7. The main difference is that you can put two quads in this setup. You can only have one quad i7. I'm just wondering if the performance would be any better with dual Xeon Nehalems compared to one i7. Is there any memory advantage? I'd like to personally have everything on one machine (MIR and Cubase 5 on a PC).

[cm]

 i7 has 3 memory channels, 2 x XEON 55xx have 6

3 memory modules on an i7 board (X58 chipset) run at 1333, 6 memory modules  run at 1066 MHz,

6 memory modules on a dual XEON 55xx board (i5520 chipset) run at 1333 MHz, though 12 again drop to 1066 MHz (i know, strange rule, but thats how it is currently designed by intel)

if memory can be unbuffered, needs ECC or even ECC fully buffered and for which ranking can be used please refer to the specs of the respective motherboard (i would recommend to strongly stick to the compatibility list - such machines can be very sensitive)

 

i am expecting ~175% power for the dual quad XEON 55xx compared to an i7 running at the same frequency - published benchmarks with databases (HP, Dell) indicate this is not only marketing speech ...

christian

[DG]

The dual quads now seem to be shipping from various DAW builders, and at a pretty good price. Looking at the new Mac Pros, the PCs are coming out much cheaper, and with a better spec.

I think that in a few months time they will be very good value for money, and whilst they may not be the best option for applications like Cubase (which is not very well coded for more than 4 cores), assuming that MIR is designed to utilise all cores efficiently, these machines will be great. Even more so when the 6 and 8 core Xeons are released at the end of this year. [8-|]

DG

[Austin]

The biggest benefit of the PC over the Mac Pro is the fact that a 5500 Series based PC can carry 12 dimm slots compared to the Mac Pro's which only has 8 dimm slots and because of the tripple memory architecture which you have to use ram sticks in 3's. This makes a huge difference as far as expandability. I ordered an Octo Xeon 2.66 with 12 GB of ram for now with the option of going all the way to 24 GB in the future :). This should be one fast machine!