Slashdot Mirror
US Bans Loud Commercials
bs0d3 writes "On Tuesday, the FCC passed the Commercial Advertisement Loudness Mitigation Act, or CALM. It's a law that states all commercials must run at the same volume as network newscasts. The same applies to network promos. The responsibility falls on cable providers like Comcast or charter. The law will not take effect until next year which leaves it plenty of time to be challenged in court by cable providers or advertisers."
there's standards for loudness in most countries, but they're completely ignored by the broadcasters. they take an ad that's the correct standard volume and go ahead and turn it up anyway.
Here is a link to the FCC website for the actual text of the Report and Order regarding implementation of the CALM Act.
Too late when they've blown your speakers or woke your neighbors up.
Check your premises.
The FCC is implementing a law passed by Congress. The FCC did not "pass" anything.
Dog is my co-pilot.
Wrong. Congress passed the 2010 Commercial Advertisement Loudness Mitigation (CALM) Act and it was signed by the President on December 15, 2010, a year ago tomorrow.
What the FCC did yesterday was to adopt rules restricting loud commercials, as it was required to under the CALM Act, which will become effective one year after adoption, on December 13, 2012.
All of the comments so far are "it's about time," so I'm just going to respond to the first post I see.
There already was a standard requiring commercials to limit loudness. A commercial could not be louder than the program it was accompanying, which meant it could not be louder than the loudest point in the programming. What that meant is if there was a single gunshot in an hour, your commercials in that hour could be very, very loud. Also, loudness was not weighted. High-pitched ringing and speaking at the same level were considered equally loud, even though human hearing is skewed (A-weighting) to perceive speech as inherently louder.
So what this really does is 1) re-define what constitutes "loud", and 2) give the process some teeth.
The rule is based on ATSC A/85 RP (70 page PDF), which most definitely is not just a simple amplitude definition.
It's loudness as defined by the measurement technique in ITU BS.1770, which is a lot more than amplitude.
Since this is Slashdot, I'll share some details on the problem of measuring loudness.
Loudness is difficult to measure objectively, because loudness is what a human experiences when listening to audio. Intensity, on the other hand, is easy to measure; just get a sound level meter.
Why is loudness different than intensity? Because the human auditory system contains a natural filterbank that divides incoming audio up into multiple bands, and then applies an exponential scaling function to each band. Old books and papers call these bands critical bands; I think the more modern concept is ERBs.
For sounds that hit only one band, such as a pure sine tone, the intensity of the sound is a good approximation of loudness. But sounds that hit multiple bands scale roughly linearly in the number of bands hit. I'll give an example.
If you generate a pure sine tone at power level X, and then generate two sine tones each at power level X/2, then the measured intensity will be identical. However, if the two sine tones are in different bands, the loudness will be nearly double.
So, as a rule of thumb, the more frequency bands a given sound hits, the louder it is at any given power level. Something that sounds like white noise will be louder than something that sounds like a clear bell tone or a single flute note.
The people who make commercials know how to game the system. I'm pretty sure that there were already limits on measured intensity of commercials, but that wasn't enough to solve the problem.
Imagine you are driving along, listening to a radio show. Maybe talk radio, maybe NPR, whatever. You have the "volume control" knob on your car radio set to a comfortable listening level. The radio show only has audio at typical human speech frequencies, and isn't trying to sound loud. Now comes the commercial, which smears its audio all over the spectrum; it puts processing on the voice, with reverb and stuff. "Sunday Sunday Sunday-y-y-y!!!! M-m-monster truck demolition derby!!!" or whatever. It's not your imagination, it really is louder. But a sound level meter might say it's the same as the radio show content, or only slightly higher intensity level.
The company for which I work (DTS) has a solution to the problem called "Neural Loudness Control", and there is a white paper available that really goes into detail about this stuff, so you don't need to stop with my lame explanation. NLC has a full "loudness model" that approximates the human auditory system when computing a loudness metric; but it also can operate in a mode that follows the new standard.
Also, here's a PowerPoint presentation by JJ Johnston about loudness vs. intensity.
So the new standard, 1770, is a pretty easy-to-calculate approximation of loudness. You apply two filters: one that simulates the transfer function of an average human head, and the "RLB weighting curve"; then compute mean-square energy on the result. This is simple enough that nobody really has an excuse in the 21st Century that it would be hard to comply.
I'm a little worried that it is too simple, and there might be ways to trick it. For example, it doesn't seem to handle audio that is smeared across multiple bands to make it sound louder. But I'm not actually working in the area of loudness measurement, and from what I've heard, 1770 works okay for most stuff. It's better than no standard.
And on the gripping hand, 1770 is the law now.
steveha
lf(1): it's like ls(1) but sorts filenames by extension, tersely
The bigger problem is working around the requirement by (ab)using the principles of the psycho-acoustic modelling of sound, like with A-weighting and equal-loudness contours.
http://en.wikipedia.org/wiki/Equal-loudness_contours
Essentially, the human ear's perceived intensity at different volumes is frequency dependent. One trick is using an auto-tuner to "shift" audio to nearby frequencies so that the overall loudness (as measured by an ideal microphone) is within the acceptable limits in the proposal, but the human ear "hears" them as if they are louder than they really are.
WHAT?! WHY WOULd YOU WANT TO DO THAt?!
FTFY.
Easier than you think.
Go open any audio-editing software there is a filter called "Normalize", it used to be standard in mp3/cd ripping software.
All the Cable co's have to do is pre-normalize the commercials to the program being played, they can do this by just keeping a running loudness average and normalizing on the fly. It's so easy you wonder why they haven't been required to do it before now.
Now's actually the best time because as more analog gear is taken out of the delivery stream, the less excuses they have for not doing it. Before you could at least claim that normalizing the audio wasn't possible because they're played off of tapes that would take time. But this hasn't really been true since 1997 or so when the C-band satellite systems started switching to all digital, and the "pizza dish" DTH became the first all-digital path into the cable plant (1994 was when DirectTV became available, but the uplinks were still going through analog paths.)