|
Mixing Matters:
Those Annoying Volume Jumps!
|
|
I want to talk about a big problem faced by the broadcast sound business, of large volume jumps while viewing. If you have ever watched television you know that volume jumps are everywhere. You can barely lift your finger away from the remote control before something is too loud or too soft again. It is such a pervasive problem that Washington Congressmen, who surely have more important concerns, have discussed regulating television volume bumps. No one likes it the way it is now. Or do they?
A typical broadcast distribution chain has many stops where the audio’s volume can be altered, sometimes accidentally, before it reaches the viewer. This might explain why one cable channel is lower overall than another. Sometimes there is a big difference between programs like a local news program vs. nationally-distributed syndicated shows. Sometimes though a program contains volume jumps within itself, like from quiet or intimate dialog to explosions and pounding music. Feature films that are not re-mixed specifically for broadcast usually have such troubles.
Enter Quality Control
Broadcasters have recently reacted to widespread viewer complaints about uneven volume jumps by enacting stricter controls on the way programs are mixed. Mixes that are destined for broadcast are put through technical tests called “Quality Control” or “QC”. Each broadcaster has a slightly different idea about how “loud” and “soft” a program should be and has tried to define those ideas as meter-measurable quantities. In other words, each broadcaster wants to see its meters move by specified amounts. The meters are used to conduct tests on each new program prior to its first broadcast, to make sure it meets the broadcaster’s technical specifications. Those of you familiar with Quality Control know that it can generally trump artistic or perceptual choices. If a sound mix flunks a QC test, the program is sent back until the problem is fixed, i.e. the meters move just the right amount. If it passes then the mix must be GOOD! We therefore have a responsibility to make certain our sound mixes move the broadcaster’s QC meters by the exact amounts that are specified. This part of our mixing job has taken on increased importance over the last few years, some would say it has become onerous, and yet volume jumps are still a big source of complaints.
All of this effort has not really worked and the reason is obvious. I hope you will forgive me as I rant for just a moment. We sound people already have mono, stereo, and various surround speaker formats to contend with simultaneously. Broadcasters impose their own unique and regularly updated technical specifications. Within the last decade we have seen our list of deliverables (mixes and sub-mixes given to the broadcasters) increase from the two stereo mixes of days gone by to 6-8 or more items that spread across 12 to 25+ audio channels. We routinely deliver “mono compatible” stereo mixes, surround mixes, and “split” mixes in various iterations with various QC specifications. A primary purpose of these specifications is to prevent volume jumps and yet viewer complaints have not subsided. What is wrong?
Nudge Nudge Wink Wink
Come on, we all know the real deal. Advertisements are the most egregious offenders BY FAR in the game of volume jumping. In the US and indeed the whole world, it is normal for advertisements to be made as loudly as possible, or at least as loudly as can be gotten away with, in stark contrast to (and in defiance of) QC constraints that are placed on the actual programs. In other words, they do not play by the same rules. It may defy logic that commercials are not subjected to the same rigorous specifications as the programs. They are supposed to be but they are not. Lip service is paid and claim is made that ads meet all specifications but what we really have is a “nudge-nudge-wink-wink” system that routinely accepts ad mixes that, to start with, contain peak sound levels 2 to 4 db greater than the programs are allowed to have. (For those who do not know, a “db” is a standard unit measure of audio power. A change of 3 db is considered to be a moderate change.) To be fair, “peak” levels, by themselves, are not indicators of high perceived volume but they are a contributor. We will look at another factor in a moment. My point now is that it is routine for broadcasters to accept advertising sound mixes that break their own rules, rules that were designed to ensure consistent sound levels for viewers. Advertisers pay to place their ads on broadcast outlets and I suspect struggling broadcasters do not want to offend their benefactors. Also, some broadcasters consistently have much louder commercial blocks compared to the programs they are running. The exact reason for that is not clear to me, but it is perfectly clear that the system is willfully broken.
Great Britain is barely ahead of the US in this regard. The U.K. Advertising Standards Authority, in 2008, proposed new measurement guidelines specifically for advertisements, to reduce the perceived volume differences between ads and programs. Under these guidelines, the measurement techniques for advertisements are not identical to those for programs, but they are intended to make viewer perceptions of volume roughly equal. It is a relatively new initiative and the jury is still out as to whether it will be successful. I hope it is.
Enter Dialnorm and LKFS
Perceived volume is what we really need to measure and control. “Peak db” is something that meters can read easily. The two have no direct relationship. Can a meter really measure perceived volume? The truth is we do not yet have a totally effective method. Human hearing is very fickle indeed, but manufacturers are trying to find more reliable measurement techniques. Several years ago a new concept of measuring loudness was introduced. Dialnorm is supposed to show how loud we perceive a program’s dialog to be, derived from several aspects of a program’s actual dialog content. It might seem odd that only the dialog, NOT music and sound effects, are measured. The idea is based on an assumption that dialog, not music or sound effects, is what is most important to a viewer who is adjusting his volume control. A similar and more recent attempt, the "LKFS" measurement (defined in ITU-R BS.1770-1), has proven to be just slightly better at predicting human perception. Both are in use today.
Dialnorm and LKFS, "loudness" measurements, are not perfect and can give readings that are contrary to what we actually perceive. Nonetheless, one or the other is the latest measurement of choice by QC departments everywhere and MUST be satisfied. Broadcasters usually specify their own narrow range of “short term” and “long term” loudness variations that they will accept. This is to allow some for some natural variations of speech while maintaining tight control of overall volume. In all cases I have seen, soft dialog is strongly discouraged, at least according to the loudness numbers on the specification sheets. In practice, though, some broadcasters seem happy to relax their QC vigilance for some of their shows so that dialog is allowed to be a bit softer overall, allowing some room for bigger music and sound effects crashes. Remember that Dialnorm was conceived to measure the dialog volume only. Right or wrong, Dialnorm readings generally ignore music and sound effects volume bumps and so do not directly address one aspect of viewer complaints. When you put a show with relaxed (softer) Dialnorm measurements back to back with “louder than loud” advertisements, you can quickly understand why complaints have not been abated. LKFS measurements are sometimes made of only the dialog content, sometimes of the overall mix, but if you study either loudness reading during an evening of television viewing you will see that a typical commercial block tends to measure MUCH louder than the programs, 4 to 6 db above what the “rules” would seem to allow. Dialnorm and LKFS are not perfect measurements but they show us what we already know. Commercials are a lot louder.
Annoyance Factor
In the feature film world there is a rough equivalent to Dialnorm or LKFS measurement that has been imposed on trailers, that run just prior to the feature presentation. The Trailer Audio Standards Association uses a measurement of something it calls “annoyance factor”, otherwise known as “82 db leq(m)”.The idea is that trailers should not be louder than the star attraction. Unlike Dialnorm, the “annoyance factor” is evaluated for the whole sound mix, not just dialog, and has generally worked well for many years. Its success is why I applaud Great Britain’s attempt to reign in on advertisements.
When we mix broadcast programs we often have to work as hard to please the meters as we do to please our senses. My hope is that all of our efforts to meet broadcaster’s specifications will not forever be wasted by the same broadcasters who drown us in loud advertisements. Loud advertisements are a problem created by advertiser's demands upon the broadcasters. They encourange viewers to fight back with their mute buttons. Maybe the problems will disappear entirely as paid advertising copy is merged into programs and news broadcasts. Maybe television programming will just become one long commercial. Until then, isn’t there some way to keep everyone happy?
|
| Mixing Matters is an open-ended series of short, easy-to-digest articles about some of the grey corners in our world of sound mixing for film and television.
©2009 Richard Fairbanks
Do not quote or copy without prior written consent
|
|
|
Click to return