Good Article on Sound Mixes

[William Kucharski]

Look, it doesn't help anyone if we keep denying this. Research has shown that hearing loss can already start at prolonged exposure to 85 dBA. OSHA regulations, to be honest, are a big joke. They allow what? An exposure of 100 dBA for 2 hours in an 8 hour time period? At those levels, a constant exposure of 15 minutes or longer can already cause permanent hearing damage. Also, cinema is supposed to be entertainment for the paying customer, not work.

Governments all over the planet start noticing this and are actively looking at far stricter rules, I for one, are afraid that those rules will be so limiting, that we eventually can't have GREAT sound anymore... I for one, love explosions to be loud, but I don't want to sit through 2 hours of constant explosions.

Then I'm done going to movie theaters and will only watch at home, problem solved.

Well, this is a topic about sound mixes, so why not discuss this right here? I honestly think Nolan's ideas about sound mixes are broken and somewhere between Inception and The Dark Knight Rises, he must have damaged his own hearing.

We've played Interstellar at 7 in our own screening room and some of the scenes are just obscene. If people start sticking their fingers in their ears, you know stuff isn't right... And yeah, that room gets re-calibrated a few times a year, unlike your average multiplex.

Thankfully I will never need to patronize your theater.

If I'm watching at home and things aren't falling off the walls at the rocket launch in Interstellar, it's way too quiet.

If a customer is sticking their fingers in their ears I wouldn't want them as a customer.

In the end, Nolan must remember that he serves his customers, who are those who pay money to see his films. Many people, in fact, most people I speak to, dislike the fact that he allows dialog to be completely muffled. Most people do think that his sound mixes are far too loud and quite a few do avoid his movies in theaters. Theaters, meanwhile struggle with those complaints. While many still want to play those movies as intended, their audience simply isn't buying it, so they take matters in their own hands.

Then let them. I'll look for theaters that give a damn.

Well, we had those very same people asking us to "turn it down a notch"...

The proper answer is one I've told people several times - "complain to the studios, this is the volume it was designed to be played at."

It's becoming more clear I need to only watch at home to see movies properly presented for a variety of reasons, pity I can't match the screen size.

If you can't tell, I feel quite strongly about this, but the majority get their way which is why AMC and Regal are predominant in the US.

[Marcel Birgelen]

Thankfully I will never need to patronize your theater.

If I'm watching at home and things aren't falling off the walls at the rocket launch in Interstellar, it's way too quiet.

If a customer is sticking their fingers in their ears I wouldn't want them as a customer.

I don't run a theater. We do have a screening room though, one who is both calibrated by ourselves and an external, independent company...

What are you watching at home? The 70mm print? The DCP release? You think the mix on that BluRay is the same?

And the rocket launch isn't even the scene with the biggest issue.

The proper answer is one I've told people several times - "complain to the studios, this is the volume it was designed to be played at."

People aren't going to complain to the studios and neither should you expect them to do so. Most people aren't even aware of studios being a thing. Most people don't even complain to their cinema, they just stay away.

[Carsten Kurz]

Again, 7 is not 7. Professionals know this for a long time.


https://internationalcinematechnolog...Vessa-Sony.pdf
https://f.hubspotusercontent00.net/h...r0994-2014.pdf
http://www.film-tech.com/ubb/f16/t002075.html
https://usir.salford.ac.uk/id/eprint...is%20Final.pdf

Calibrating will only get you somewhat closer than not calibrating. Also, you can only calibrate equipment, not your patrons ears, habits, biography, expectations.


We had a few complaints during our screenings of the recent Dune and James Bond (even one during 'The Power Of The Dog', which is just a talkie.). And we played neither of these AT 7. Just loud.
I always try to talk reasonably to these people. I believe they are suffering.
Actually, after these recent discussions, I bought a bag of these on ebay:

https://www.honeywellsafety.com/SKU/....aspx?site=/de


and placed them at the counter, so if staff encounters complaints, especially during action movies, they should handout a pair for free. They only cost us a few cents and I consider it a service to those people that have a different sensitivity, but should none-the-less be granted a movie viewing and hearing experience.

- Carsten

[Steve Guttag]

Again, 7 is not 7. Professionals know this for a long time.

No...those that don't like the X-curve spew it because they feel they have a better way...that has yet to become standard.

7 is 7 (or 0dB, for those misguided manufacturers that use a dB scale for a volume control). Your links don't even dispute 7 is 7. I particularly like the one to Film-Tech where I gave a lengthy (and I stand by) discussion of SMPTE 202.

"7" like 85dBc are single number representations of complete responses. The same goes for 14fL (48cd/m^2). 85dBc doesn't tell you the tone, distortion, the reverberation...etc...and 14fL doesn't tell you the color, the uniformity...etc.. When single number representations are used, they cannot, in most cases tell the whole story. They are often used as a single point of pass/no-pass. Furthermore, all such specifications have qualifications put on them. For instance, 7 represents 85dBc. That "c" has meaning to it all. It specifies the frequency range of interest and what can contribute to the single number. If the response of your system does not fill that range, then the number will be invalid. Just like when you use a silver screen with a 2.5 gain, 14fL is meaningless because the distribution is not >85%. Numbers do not live in a vacuum. They have qualifications if they are to have meaning. Furthermore, 85dBc has a tolerance (which I'd have to look up). I have often questioned why techs will calibrate their system using expensive (and hopefully calibrated) microphones and then use a cheese-ball Radio Shack meter as their SPL meter. One has no clue as to the frequency response of that meter and if it is more sensitive to a frequency your tuning may be a little high or low on so it gets exaggerated. The SPL meter on the RTA is sufficient and, at least, you can see the frequency response it is basing its number on. Furthermore, the actual response should be the guide to the level as the SPL is just the logarithmic average of the discrete bands. One has more information before looking to the cheese-ball hand held (e.g. inconsistent) meter. It is like reading a circuit with your Ohm meter and then, after seeing near zero ohms switching to a continuity tester.

Where 7 can be skewed from 7 can be if the pink noise source is not consistent I believe one with a 12dB crest factor is the cinema standard. That isn't the fault of "7" but of the test apparatus if a cinema processor chooses to do their own thing. But that's just it. One thing cinemas have enjoyed for a long time is a degree of uniformity from dub-stage to cinema. No, they are not identical but compared to other forms of sound reproduction, we have it best.

I've done a LOT of studio screenings in my time (did one less than a month ago, even) and 7 was indeed 7. Furthermore, the room I was in had a different studio screening just before mine so I was measuring behind someone else's work using completely different equipment and mic positions. 7 was still 7. The degree of agreement between two different techs using two different pieces of equipment was quite impressive, in my opinion. In fact, I could tell which speakers s/he spent more time on (it was quite evident to me). But, if you had pulled out a calibrated meter (with proper frequency response) on either of us using either the processor or played a pink noise DCP...you'd have measured essentially the same level to within the accuracy of where one placed the microphone relative to the screen. I can tell you that person's spectroradiometer was calibrated too (or we were equally off!).

Now, what 7 represents in terms of sound quality, that isn't definable in a single number. Too effin loud, sure (for most features...not so for those that make good mixes...I refer any and everyone to Larry Blake).

[Carsten Kurz]

Steve, if we had trouble maintaining consistent loudness same location same equipment, we were in real big trouble, no? That's not what we are talking about here.

[Harold Hallikainen]

Since some articles on the X-curve were posted, I'll throw in my thoughts.

I see the X-curve as a pre-emphasis / de-emphasis system. In the theater, de-emphasis is caused by the high frequency attenuation of the perforated screen. In the SMPTE tests, I was there for tests on a dub stage. That room had a woven screen. HF rolloff had to be applied in the sound processor to comply with the X-curve. This, to me, is still just deemphasis.

People mixing movies say they do not purposely boost the highs to pre-emphasize the audio. But, to me, this means that they are mixing the movie to sound bad since they are listening on a dub stage with deemphasis (either a perforated screen or electrical deemphasis).

One of the papers posted shows a typical audience prefers a different curve. But, since an auditorium equalized to be close to the X-curve is similar to what the sound mixer heard, are we just saying the audience wants to hear something different than the mixer created? I think this says nothing about the validity of the X-curve.

There have been proposals to make cinema mixes flat. This would then require HF boost at the theater to compensate for screen HF attenuation. The only difference between this and the existing condition is where the HF boost is applied (before or after recording).

I see nothing wrong with pre-emphasis de-emphasis. It is used in magnetic recording, optical recording, phonograph recording, etc. Since we have a lot of content made with this, why not keep it? If we want, we could define a simple electrical pre-emphasis that duplicates the X-curve. The dub stage with a woven screen would "mix flat" and the pre-emphasis applied just before recording. A dub stage with a perforated screen would have a screen compensation equalizer that results in flat sound. The mix would then be flat and pre-emphasis applied just before recording. At the cinema, electrical de-emphasis could be applied (giving flat sound), then screen correction applied. Or, the screen could provide the deemphasis with slight equalization to make the HF rolloff of the screen match the standard.

So, there are several SMPTE members that disagree with this, but this is what I see going on with frequency response in cinema.

Harold

[Steve Guttag]

Steve, if we had trouble maintaining consistent loudness same location same equipment, we were in real big trouble, no? That's not what we are talking about here.

No! We don't have that problem. I can have different equipment show the same level in that location. But, as I said above, SPL does not tell as detailed a story as an RTA let alone looking at time-domain and THD...etc. The idea behind the X-curve and SMPTE202 really comes from measurements in real theatres and having a set of instructions, if followed using reasonably obtainable equipment, 10 different techs will end up with rather similar results. They won't be identical as some will look at adjacent bands and how they interact while others will behave like early auto-eq and just blindly adjust EQ without regard to what one is doing in the time-domain. SMPTE202 is not a maximum, it is, really, a minimum. Nothing stops a person or equipment from improving on the results however, the improved version should also fit within SMPTE202.

As for various curves and "flat." My recording engineer friends do not EQ for "flat" They don't have the 3-hinge curve of the X-curve but it is a steadily decreasing line from bass to treble. I think 20KHz is something like 3dB below 20Hz or something like that (it might be steeper but it is a constant slope). The X-Curve is similar though...it just has a flat response from 100-2KHz rather than a decaying one. Then we hinge down from 2KHz - 8KHz. That is not a function of the screen as the screen is not having too much play there. There is a second hinge above 8KHz and that is a combination of the screen and the reality that the compression drivers of the day (Altec 288, being closest to a world-wide standard) really would struggle above 16KHz, even without a screen. Even later one when JBL's 2445 (THX in cinema initial era) and then the 2446 10-years after that...being large, high-mass diaphragms still would have problems above 16KHz. So, you have that extra decay from 8KHz to 16KHz due to screen losses and the inability of drivers of the day to really reproduce the frequencies. You get diminishing returns, given the hearing losses of most people above 16KHz and, also, when the X-Curve was developed, the soundtracks of the day were being played by optical sound systems that couldn't really reproduce up to 16KHz or magnetic tracks (composite prints) that are questionable above 16KHz too.

I have zero issue with improved tuning methods. However, in order to make it a standard, and therefore expect people in the field to follow the standard to attain it, it has to be well documented, only require readily available equipment and be reasonably followed by the typical field technician. I haven't seen that yet. I've seen auto-tuning systems that have gotten much better (e.g. Trinnov or Dolby's DAD system)...when done, they do follow SMPTE202M...which you really have to do if you don't want multiple tuning presets based on how a movie was made...which I think is very impractical, at this stage. Perhaps if metadata was used, codified and guaranteed to be present or compensated for (e.g., if the server doesn't get a response from the sound system that such a tuning exists, it applies its own deemphasis to have it work with an X-Curve theatre.

Thus far, I haven't seen/heard such a presentation that could replace SMPTE202, at this time.

[Harold Hallikainen]

Thanks for the comments! See https://www.acousticdirections.com/w...62.pdf#page=15

As shown in Fig. 34, when the on-axis frequency response is smoothed over a third octave bandwidth, the loudspeaker’s frequency response falls in the tolerance bands of the SMPTE 202:2010 X-curve standard when the theater-perf and mini-perf screens are used. We conclude that the screen itself is one of the greatest contributors to the frequency response observed in a cinema.

Harold

[Steve Guttag]

Harold, do you know of anyone that watches movies at 6, 12 and 18 inches from the screen? The X-Curve also having a similar type roll off to the attenuation of the screen is coincidence. Did they measure the speaker response in a typical theatre? Say with a 60 foot deep room and put their mic at 40 feet? What sort of response do you get then? It's all well and good to measure the attenuation due to the screen type/perf type as that does add useful information but it does not tell the whole story. I definitely wouldn't assign a relationship between screen loss and the roll off built into the X-Curve. It's cooked in there but so is the room size.

[Harold Hallikainen]

https://f.hubspotusercontent00.net/h...14.pdf#page=87

The responses of the close-field microphones and the averaged close-field response of the reference cinema are far from flat. In general, the responses of the close field microphones above 1 kHz show strong similarity with both the reference position and average responses. There is a little more high frequency energy in the close field responses, which is consistent with the loss due to air-absorption as the sound travels to the audience area. In other words, the close-field microphones essentially show an X-curve with a little more high frequency response, corroborating the results shown in (2).

e) One component of the change in frequency response described in (8), apparently due to the effect of the reverberant field clearly does not occur here. There is nothing to suggest that the X-curve as measured in the room is a measurement artifact resulting from reverberant buildup; i.e. an artifact resulting from steady-state measurements of typical cinema acoustics.

[Marcel Birgelen]

The very concept of the X-Curve itself may be outdated, since it was created in a time with vastly different equipment. Although it has worked remarkably well over the many years it has been used and I don't see an easy way to get rid of it, since both recording and reproduction "stages" usually have been tuned according to SMPTE 202. As soon as a movie is recorded with other profiles, we need to tune rooms according to two or more different profiles, which is simply not going to happen.

I have zero issue with improved tuning methods. However, in order to make it a standard, and therefore expect people in the field to follow the standard to attain it, it has to be well documented, only require readily available equipment and be reasonably followed by the typical field technician. I haven't seen that yet. I've seen auto-tuning systems that have gotten much better (e.g. Trinnov or Dolby's DAD system)...when done, they do follow SMPTE202M...which you really have to do if you don't want multiple tuning presets based on how a movie was made...which I think is very impractical, at this stage. Perhaps if metadata was used, codified and guaranteed to be present or compensated for (e.g., if the server doesn't get a response from the sound system that such a tuning exists, it applies its own deemphasis to have it work with an X-Curve theatre.

Well, this got me thinking a bit, because you could expand the current audio specifications with optional metadata, like an "expected tuning profile" or whatever you want to call it. Combine this with the auto-tuning capabilities of many modern decoders and you could do auto-tuning based on the recorded preferences of the room and the "expected profile" on-the-fly. Still, you'd always need an SMTP202-conforming mix for all those theaters that don't support this feature.

I guess the optimizations this allows for, in the end, aren't worth the efforts though as this has very little "marquee value". It's easier to kick up the main fader a notch and call it a "Deluxe" screening, at least that's something most people will actually notice...

[Carsten Kurz]

It would be rather easy nowadays to use a software defined approach for tuning as a standard - that is, a microphone/multiplexer, an A/D, and a software running on a notebook. However, someone has to do it, and that someone would need to be able to create the necessary business penetration for it to become a defacto standard. There are many software based solutions already for audio system tuning, but none of them has become a standard. It is more likely that the calibration methods that are built-in most current dedicated cinema processors become de-facto standard.

[Ed Gordon]

Things Easier to Hear Than Movie Dialogue

Capture7.JPG
Credit:
https://www.newyorker.com/humor/dail...movie-dialogue

By Andy Babbitz, New Yorker Magazine
February 6, 2022

[Harold Hallikainen]

I wonder if Academy voters for best sound listened to the movies at "7".

[Marcel Birgelen]

Sure, they tune their TV to "7" on the scale of euhm.... "Volume". :P