Television Commercials, Loudness, and OP 48

....a journey into the realms of fantasy

In recent times there has been a lot of discussion in the general media about the serious problem of television commercials and promos being significantly louder than adjacent programmes. Pushing aside more trivial issues such as the Middle East, War, Medicare, and The Free Trade Agreement, the issue finally reached ministerial level.

Probably as a result of this, in July 2004 the Australian Broadcasting Authority registered a revised Commercial Television Industry Code of Practice.

This code contained a section referring to the loudness of television advertisements:

1.11	Commercials must not be excessively noisy or strident.
1.12	Licensees must do everything reasonably possible to ensure that commercials do not sound louder than adjacent programming.  Studio transmission must not be increased from normal levels during program breaks.
1.13	A licensee shall be deemed to have complied with Clauses 1.11 and 1.12 provided it ensures that the relevant requirements of Free TV Australia’s Operational Practice Note on Loudness of Advertisements, as amended from time to time, have been met.  This requirement is satisfied if a person submitting a commercial certifies to the licensee that all requirements of the Operational Practice Note on Loudness of Advertisements concerning compression, limiting and equalisation have been met.

The entire Code of Practice can be downloaded at: http://www.aba.gov.au/tv/content/codes/commercial/index.htm

As a response, Free TV Australia, (formerly FACTS), has separately prepared a document called “Operational Practice 48 – Audio Levels and Loudness for Advertisements”, affectionately referred to as OP 48.  You can download it as a PDF from here, since you need a degree in I.T. to find it on their website (http://www.freetvaust.com.au).

(By the way, guys – it’s about the name – the nearest thing to free TV in Australia is the ABC, and even that costs 8 cents a day.)

The puzzling thing about OP 48 is that, although it purports to be a technical document, it contains no real technical information that would assist in addressing this “problem”.

Apart from a section which insists that 1Khz line-up tones should be at –20dB relative to clipping, there are no technical specifations outlined at all. There is no specification of peak level, which would seem to be more relevant, since a programme with peaks of more than 10dB above reference is usually illegal, although apparently compliant with OP48.

Most Australian production studios use AES-EBU standards, which specify a line-up level of –18dB, with peaks 8-10dB above that, although -20dB is quoted as the SMPTE standard, and is also frequently used.

The document goes on to say:

“Aggressive use of compression on components of the sound track or the final mix of a sound track must not be used for the purpose of producing  excessively noisy or strident material.”

And later:

“….limiting must not be  used for the purpose of producing excessively noisy or strident material.”

There are no guidelines as to what constitutes “aggressive” use of limiting or compression, and since these are largely subjective or creative decisions, it would be extremely difficult to quantify them, since the implementation of dynamics varies widely from case to case.

In any case, if the sound mixer does not compress material fairly heavily for commercial television, whether it be for programmes or commercials, the network will do it for us, since the one thing they hate is wide dynamic range.

In regard to equalisation, OP 48 states: 

“5.3   Spectral Manipulation (Equalisation)    Alteration of the spectral response of the audio system to emphasize  those frequencies to which the ear is most sensitive, must not be used for  the purpose of producing excessively noisy or strident material.  While equalisation is a basic tool and often used in audio production, care  should be taken to avoid excessive amounts of equalisation that could  cause overloading of broadcast audio chains.”

Most sound mixers use equalisation for a wide range of purposes, one of which is to tailor audio for the hopelessly inadequate speakers fitted to most television receivers, and to allow for the often poor listening environment at home. This makes the heavy use of both equalisation and dynamics not only desirable, but essential.

The fact is, commercials are not too loud, but rather, programmes are often not loud enough. It is easy during the course of a thirty-second commercial to keep the average energy high. With a close-miked voice-over and a pumping music track, all you have to do is peak to –10dB  to achieve a loud, present track in the domestic environment.

It is more difficult – and clearly undesirable – to do the same thing with a one-hour programme. Programmes have light and shade, and when the network jumps without warning from a quiet drama scene to a commercial as described above, the effect is bound to be jarring. This is hardly the fault of the sound mixer or the producer of the commercial.

Perhaps networks should pay more attention to how and when they insert commercials and promos.

Interestingly, OP 48 puts the responsibility for this back onto the production company making the commercials, even though it is clearly beyond their control:

"Producers will be required to certify that their commercials comply with this  OP as a condition of acceptance for broadcast. This will require a  declaration of compliance when the commercial is submitted to the  broadcaster.  The Visual Identification between the alignment signals and the  countdown must include an additional field indicating compliance with this  Operational Practice i.e. Compliant with OP48 (Audio Levels and Loudness).”“

This is a worrying development, for a number of reasons.

The ID board is nearly always put on the tape before the sound mix is done, because of the requirements of the workflow. So the producer must assume that his sound mix will conform, since he is required to declare, ahead of the event, that it does.

There is, in any case, no way of measuring whether or not a spot conforms to OP 48, even after the sound mix is done.

To maintain consistent loudness, the producer would have to know whether the spot is going to be inserted into a football match or an art film, as the effect will be completely different. To play it safe, and be sure of complying, all commercials would have to be mixed at least 6-8dB lower than programmes. This would mean that, in some cases, they would be too low compared to the adjacent programme. The sponsor would then be entitled to complain that he was getting less than his money’s worth.

(The Sponsor – remember him? He’s the guy who underwrites our “free” TV, and then charges us for it, by building it in to the price of everything, whether we watch it or not.)

This is, of necessity, clearly the responsibility of the broadcaster, and OP 48 would appear to be just an attempt to deflect complaints from their viewers.

(The Viewers - remember them? – they’re the ones who pay for “free” TV – see above.)

The problem may become academic with the eventual demise of analogue TV, some time after 2008. The Government has promised a review then to decide whether the existing analogue networks should be switched off, leaving us with a digital-only system.

Digital television in Australia (DVB) supports both MPEG audio and Dolby Digital (AC3).  All networks use MPEG, but only some use Dolby Digital.

Dolby Digital, which can be anything from 2.0 to 5.1, has provision for a metadata facility called Dialogue Normalization, or DIALNORM. This is simply a way of standardising levels between adjacent programmes. It is not a form of compression, and in fact does not affect the transmitted audio at all, since it is applied in the receiver.

DIALNORM uses the fact that we humans intuitively know how loud dialogue (speech) should sound. All Dolby Digital programmes carry a DIALNORM number, which tells the receiver how much attenuation to apply in order to achieve a standard level. The default DIALNORM setting is –31, which means a programme with a DIALNORM of –27 will be attenuated by 4dB. This is a very common setting for drama programmes. A compressed, energetic commercial or music programme may have a DIALNORM of, say –20, causing it to be attenuated by 7dB compared to an adjacent drama programme, or 11dB against the default. Very loud programmes may have even higher DIALNORMS.

The DIALNORM can be set in the mixing studio, or at the time of restripe, if the programme is encoded with Dolby E. Because it is dialogue-related, it allows other elements to be louder or softer, as per the original mix. For example, an action movie with a DIALNORM of –27 can still have sound effects and music peaking to 0 (27dB higher!) This will be great news to viewers with home theatre systems, but may not work so well on a portable. But it does mean that the dialogue levels should be consistent between programme and commercials, even if the programme is in 5.1.

It is not clear what will happen if some networks continue to use MPEG audio only, which does not support a DIALNORM feature. It may mean that some compression will still be applied at transmission, and programme level inconsistencies will still occur, particularly since the current rule about leaving 10dB of headroom above peak programme level need no longer apply. (This is primarily a requirement of the analogue system, to avoid audio peaks getting into the video during transmission and causing “soundbars”). Networks have traditionally applied compression and limiting in the transmission chain for this purpose. In theory, they can (and should) stop doing this in digital, particularly if a standardised dialogue level can be achieved.

In the meantime, if you mix commercials, don’t be surprised if you or your client are asked to sign a document stating that your mix is not too loud, too soft, too round, or too orange…. And you will be much too polite to suggest where such a document should be lodged.....