[BC] DTV Audio Levels

[Robert Orban]

At 05:37 AM 6/15/2009, Rich Wood wrote:
>------ At 08:23 AM 6/15/2009, Cowboy wrote: -------
>
>>  This one is bigger than radio guys can ( at first ) imagine !
>>  Not only is there reluctance, but that reluctance is entirely justified.
>>  First, you have to un-embed the audio. Then, you correct the level.
>
>I don't think it's a radio vs. TV guys thing. 
>Very simply it's listener/viewer fatigue and 
>irritation. Viewers don't give a damn how it's 
>done. All that matters is that they enjoy the 
>experience and stay tuned. Years ago, when the 
>FCC got involved, complaints weren't about bad 
>pictures. They were about loud commercials. Now 
>we have extremely loud commercials, some in 
>Dolby Digital 5.1. This time the FCC probably 
>won't get involved if there's no fine attached. 
>It's not a profit center. A TV station will be 
>less of a profit center if viewing falls off or 
>a local advertiser gets no results because his spot has been zapped..

Rep. Echoo has introduced a bill in Congress to 
regulate loud commercials, which means that the 
FCC may get involved, although in fact it is 
already against FCC Rules to run excessively loud 
commercials. I recently sent her the following letter:

It has come to my attention that you have 
introduced H.R. 6209, otherwise known as the 
Commercial Advertisement Loudness Mitigation Act. 
According to the press material I saw, the bill 
would require the Federal Communications 
Commission to "prescribe a standard to preclude 
commercials from being broadcast at louder 
volumes than the program they accompany."

I am Vice President and Chief Engineer of 
Orban/CRL a company that manufactures automatic 
loudness controllers for television. Although the 
company is based in Arizona, our main R&D 
facility is here in the Bay Area, in San Leandro.

While I would ordinarily be delighted to see any 
government action that would help us sell our 
products, the fact is that your proposed bill, 
like any other attempting to proscribe loud 
commercials, must have a clear, provable 
technical metric that determines whether a given 
broadcaster is in violation. Absent this, one has 
a nightmare scenario like that currently 
surrounding "obscenity" and "indecency" 
regulations-is a broadcaster guilty or not 
guilty? Whether one is talking about porn or loud 
commercials, claims of "knowing when one sees it 
or hears it" are insufficient to make good law.

"Loudness" is not simple. Loudness is subjective: 
it is the intensity of sound as perceived by the 
ear/brain system. No simple meter provides a 
reading that correlates well to perceived 
loudness. A meter that purports to measure 
loudness must agree with a panel of human 
listeners. Moreover, listening tests have 
established that different listeners perceive loudness differently.

The literature on loudness measurement stretches 
back to at least the 1920s, with work first done 
at Bell Laboratories by Fletcher, Munson, and 
Snow. There are many different techniques that 
attempt to measure loudness with electronic 
hardware, three of which have become 
international standards. None of these three 
standards fully agrees when they are used to 
measure the loudness of material commonly broadcast.

To illustrate the complexity and difficulty of 
the question, I have enclosed a fairly recent 
research paper (from 2004) addressing the issue: 
Esben Skovenborg & Søren H. Nielsen, "Evaluation 
of Different Loudness Models with Music and 
Speech Material." This paper is technical and 
understanding it requires a background in 
electrical engineering and statistics. However, 
any intelligent nonprofessional reading it should 
get a sense of how difficult it would be to 
determine objectively if a broadcaster is in fact 
in violation of the proposed limits in H.R. 6209. 
(The 101 references at the end of the paper are 
the best collection I know of for those 
interested in fully understanding the subject.)

The broadcast industry, both in North America and 
Europe, is well aware of the problem of loudness 
consistency in broadcasts. There are many 
competing products and techniques that purport to 
deal with it. All of the major U.S. television 
networks have studied the problem and are in the 
process of implementing measures to mitigate it. 
The International Telecommunications Union (ITU-R 
WP6P SRG3) is also working on the problem.

There are two main approaches to the problem. The 
first, promulgated by Dolby Laboratories, 
includes "metadata" in digital audio streams. 
This "data about the data" indicates the level of 
normal speech in the program and provides means 
for the listener to reduce the dynamic range (the 
range between quiet and loud material) at the 
receiver if she wishes. However, the fully 
dynamic range of the original sound mix can be 
enjoyed by a different listener if she desires.

While the Dolby method is elegant and would 
appear to offer all things to all people, 
implementing it through the entire broadcast 
chain has proven to be very challenging. Ideally, 
the program originator should create the metadata 
and this should be passed, unchanged, through the 
entire broadcast chain to the listener's 
receiver. In practice, there are three main 
limitations. First, the broadcast chain is very 
long and complex, so it is not easy to retain the 
Dolby metadata all the way from a Hollywood 
mixing studio to a TV listener in the Midwest. 
Second, a considerable amount of material in a 
typical broadcast was never authored with Dolby 
metadata (such as reruns of TV shows from the 
'70s). Third, there is a temptation for 
commercial producers to game the system by 
authoring metadata that makes their commercial 
just a little bit louder than everyone else's. 
This makes it essential for broadcasters to 
review all material before it is broadcast to 
make sure that the metadata is "honest" and 
reauthor it as necessary. Dolby Labs manufactures 
equipment to do this automatically for a TV 
station's library (usually located on computer 
hard drives these days), but even if broadcasters 
purchase this equipment, it cannot be used with 
live news and sports and the broadcaster is at 
the mercy of Dolby's philosophy of loudness 
measurement (all normal spoken dialog should be 
equally loud), which does not take into account 
material like loud sound effects or music.

The second approach to solving the problem is to 
reduce the dynamic range of program material by 
capping subjective short-term loudness to a 
preset level. This is the philosophy behind my 
company's products. While this removes the large 
dynamic range that some home theater aficionados 
love, the upside is that this guarantees a 
consistent, smooth, and easy-on-the-ears 
presentation with no extra work on the part of broadcasters.

There is another, more subtle limitation to this 
approach: subjective loudness is limited to a 
preset threshold. Therefore, if the program 
material preceding a given commercial is very 
quiet (rustling leaves for example), then the 
commercial might still seem too loud. One might 
first think that the commercial should be turned 
down to a loudness lower than the threshold. But 
what of the succeeding commercials in the 
cluster? And what if the program material 
following the cluster in substantially louder 
than the cluster's final commercial? For these 
reasons, we believe that it is best to not 
attempt to understand context because there is 
usually no practical way to know the context at 
the end of the commercial break.

In my opinion, combining subtle dynamic range 
compression with capping loudness to a preset threshold works well.

My point is this: an issue that might seem simple 
to the layperson is not simple at all. Your 
perception of loudness may not be identical to 
your neighbor's perception. Moreover, loudness 
and annoyance are not identical perceptual 
constructs. (Annoyance has been studied since at 
least the mid 1960s, mostly in the context jet aircraft noise around airports.)

The broadcast industry is well aware that it has 
a problem and is working on solutions, some of 
which have already been implemented. This is an 
ideal example of "let the marketplace work" 
because TV broadcasters do not want to lose 
viewers who are irritated by excessive loudness. 
I believe that this legislation is unnecessary 
and that it will ultimately do nothing more than 
transfer more of the nation's wealth into the 
pockets of attorneys. So I urge you to let the industry sort the problem out.