Loudness and room-size was: TV-Mix for Theatre

[Frank Kruse]

BTW: I have spoken to a professor at the sound section of the berlin film school and he will try to find some graduating students that still need a topic for their thesis and suggest this very topic we have been talking about. He said it´s a very interesting topic since everyone ahs encountered it and there seems to be very little info that directly addresses this phenomenon.

It will be interesting to see results from investigation from within the film world since they have all the bells and whistles at their facility. From small edit rooms to large scale rooms with THX approval etc. They also have all the measuring machines and enough people to do some blind tests und most importantly they have the time people from the frontline of film-bizz never have :-D

It will take some time but I´ll keep you posted what came out of the project.

Frank.

[Don Barto]

So let me get this straight…for whatever reason(s), small rooms tend to reproduce pink noise in such a way that it yields a lower spl reading than one would expect, given the way those same rooms seem to reproduce actual program material at what is perceived through actual listening to be a higher than expected level with the same monitor level setting — continuous test signal low, real world audio loud. Right?

So, rather than taking a stab in the dark (no pun) by lowering the monitor levels in a smaller room so program level seems to translate better to larger rooms, maybe we just need to find a different sort of test signal that translates better between rooms of different sizes — and all use it as a final monitor level tweak after we use the pink. How about if we all just use some little clip that starts with pink noise at the prescribed level, followed by a minute of so of maybe dialog recorded at a “good” average level. If we all used the same clip, we could then whip out our RS spl meters and maybe agree on what we should all should be looking for as to dialog level. If the RS spl meter wouldn’t be an appropriate measuring device, maybe we could agree on something else. But I guess that would take all of the mystery out of this fascinating subject. Let me clarify — I'm not being sarcastic, it really IS fascinating to me.

[Branko]

...taking a stab in the dark (no pun) by lowering the monitor levels in a smaller room...

Well, lowering the level IS, obviously, the solution, but the problem is we don't know exactly why and how much... As in the case of the X-curve, experiments have led to a solution, and we know the solution. We would just like to know how to calculate it with some precision...
Branko

[Don Barto]

Yes, I agree Branko, and I also think that studying the situation with something other than a continuous signal — something that is closer in nature to the signals we generally monitor — could yeild a better understading of exactly what is going on. In the mean time, something like the noise & dialog clip idea could provide a workaround until such understanding is within everyone's grasp.

[Frank Kruse]

don branko,

the problem with your suggestion is: what is the reference in your model? You suggest voice clips to listen too. But we are talking about the difference that occurrs between a large and a small room.

What do you want to do when you only have a small room? what do you compare it to? And when is a small room a small room that doesn´t allow to callibrate to 85? And what makes the small room a small room. Or is it the fact that the speakers are set up closer to the listener? That are many variables that MIGHT affect this phenomenon.

Subjective listening will only help AFTER you have mixed something on a large scale reference and are able to play that back in a small room with a certain amount of "accoustic memory" and set a monitor level that "feels" the same.

A mixer that starts a mix with the dialogue pre-mix needs to rely on a reference monitor level to mix too because nowadays you can set your levels by trusting your ears and not your meters because headroom is a non-issue in digital film sound. There is so much headroom on a digital sound track that rather your ears will fall off than your media will over-load. But there is one thing the mixer needs to rely on: The monitor level.

frank.

[Branko]

Frank,
I didn't propose any pink noise/dialog combination for room calibration, but I definitely think that some different kind of reference material should be used, along with pink noise and some other measurement techniques.
Decades ago it was obvious that it is the room that introduces big differences in listening conditions and many attempts were made to "exclude the room influence" from the monitoring chain. Nearfield monitoring is one of good examples - the closer the speaker to the listening position, the less the room influence, the closer we get to the speaker, more we hear the speaker itself. When you listen to a program on your headphones, it is to hear the details otherwise masked by room noise or reverberation...
Coming back to the origin of the X-curve, it is amazing how an empirical method led to such a good standard that has been around for so many years. Ioan Allen describes the process of establishing the curve: in Elstree studios, they compared the sound coming from reference close-field speakers (KEFs!) and classical theatre screen speakers far away from the listening position and equalized the screen speakers to sound similar to nearfields by playing back various program materials. Only after getting them as close as possible, they measured the response at the listening position and draw the curve! Imagine! Obviously, many things have changed since and rooms today aren't equal to those 30 years ago, so the X-curve was modified and will be modified based on listening tests many times before we finally figure out the mathematical method. Somebody in earlier discussion mentioned bass traps: a full mathematical model has been developed recently, almost 40 years after they've been invented and during this time everybody was happily using them based on results obtained empirically.
The bottom line for me is: I'd be happy if someone comes up with a rule like: "drop the level by 1 dBC SPL for "X" reduction in size, referenced to "Y" room volume.
Branko

[Don Barto]

Frank, what I am proposing is to look for something more reliable than “acoustic memory.” If 3 people in 3 very differently sized mix rooms all played the same audio clip — a clip that started with pink noise recorded at a level of –20 dBfs, followed by a section of dialog recoded with, say, –16 dbfs peaks — and all set up their monitors so the pink niose played back at a SPL of 85 dB...what would the level of the dialog peaks read on each system? One would expect that dialog would read higher on the system in the smallest room, and less in the largest room. Would it then not follow that it would be wiser to set up dialog playback in the small room so it played back with the exact same level peaks as dialog playback in the larger room? And that as a result of the level attenuation on the system in the small room, -20 dbfs pink noise playing back on that system would playback at a SPL of something less than 85 dB? How much, who knows — we all need to do some testing and sharing of results — but with the correct observed data, the smaller room — or any room — could be made to reliably playback dialog at the same subjective level as a typical large room where films are generally mixed. Granted, the measurement tools needed would probably be something more ambitious than a radio Shack SPL meter — perhaps just something as simple as a suitable reference mic running into a mic preamp and on into a digital peak meter, with the mic gain set so as to achieve a –20 dbfs reading on the pink noise would do — but such data would be valuable, so long as everyone could duplicate the test setup, and have access to the test clip...and everyone else’s playback data. That’s all I got.

[cmaynes]

Don,

forgive me for commenting on your post, but what Frank is saying is all too true- large rooms absorb sound in entirely different ways that small rooms- When moving work from place to place, your experience in hearing your mix in a bigger space is something that is invaluble. I have had many discussions with people who think this can be overcome with technology, but I frankly have not seen a viable implementation yet. Perhaps as processors become faster we may have a way to accurately compensate for room size, but I do not believe that day has arrived yet. As to using standardized setup tones, the Dolby pink noise is somewhat universal for that. I see no real benefit in using something different unless it provides a great benefit. Your suggestion about the dialog clips may be useful, but each room will have its own set of resonance that may or may not give a clear picture of the response of the system, also I think the current technology we have may not have the amount of precision which will allow our ears to be fooled. In a bigger room, the dialog mix will tend to be less effected with EQ and excessive dynamics, since the bad noise in the track is FAR less apparent 50 to 75ft away from the speaker vs on a near or mid field monitor. This is really why I thought it important to bring up the time-domain issues with the sound travelling through the air- the further it travels, the less accurate the phase alignment will be-


charles maynes

[cmaynes]

Well, lowering the level IS, obviously, the solution, but the problem is we don't know exactly why and how much... As in the case of the X-curve, experiments have led to a solution, and we know the solution. We would just like to know how to calculate it with some precision...
Branko

Well-

One could follow the Dolby and THX specs for room size- that would be a start. The problem is that unless you own a true mix stage, those dimensions may or may not be easy to implement.

charles maynes

[Frank Kruse]

Sorry guys,

I just realized that i mixed two names I addressed. "Don Branko" should have been "Don barto" sorry for the confusion.

Frank.