Loudness or level?

In this week's feature the Prof looks at normalising and maximising, and how 'loudness wars' are having a negative effect on reproduced sound
Publish date:

Recently there’s been a concerted effort to shift attention away from the idea of normalising the peak level of programme material to one of normalising its loudness. Granted, this applies primarily in the broadcasting domain, but it increasingly affects the studio business as well, not least because the so-called “loudness wars” are having a detrimental effect on reproduced sound quality. There’s been a trend, particularly with the easy availability of “maximisation” or similar algorithms on digital workstations, to “slam” mixes so that every track peaks at the maximum recording level in the digital domain. Not only is this unnecessary when digital systems have so much dynamic range these days, it can even result in increased peak distortion when such signals are MP3 encoded or replayed through some oversampling D/A convertors. Peaking the mix at least a few dB below full scale helps considerably, and greater listener enjoyment can result from a more varied dynamic structure in songs or albums that allows them the space to “breathe”. Compression to make mixes sound uniformly loud makes them very tiring to listen to. (This is not just an old fart talking – research conducted in the US showed that listeners were inclined to spend less time listening to radio stations with excessive loudness maximisation.) At a recent AES UK workshop, chaired by Crispin Murray of Metropolis Studios, mastering engineers bemoaned the pressure from record company executives to slam mixes. Masters would be routinely returned if the engineer attempted to use lower than maximum levels, or reduced the amount of compression to give a mix more dynamic range and higher quality. An interesting side-effect of slammed mixes, though, is that broadcast dynamic range processing such as Optimod can sometimes act to make them quieter than mixes that have lower peak levels. Perhaps this is the only thing, the AES workshop panel suggested, that might make the record companies sit up and listen. It seems to be pretty much a fait accompli that broadcasters will move towards the active management of loudness from now on, as the variation between programmes has become the greatest source of listener complaints. Writing in the EBU Journal, Florian Camerer – chair of a European group working to normalise the management of loudness in broadcasting – suggests that active loudness control of programme material “punishes overcompressed signals and thus automatically encourages production people to think about other, more dynamic and creative ways to make an impact with their programme”. To those like me, brought up on metering using peak programme meters or VUs, the idea of metering loudness seems unusual. It does, however, seem increasingly sensible to adopt something like the EBU recommendation to control the long-term loudness of mixes to a loudness level set at –23 LUFS (loudness units with respect to full scale). The true peak level (measured using an oversampling meter, to catch the possible true peaks that lie between samples) is then expected not to exceed –1dB with respect to full scale. DK Technologies, for example, has recently released an update for some of its metering displays that enables them to display the new ITU, ATSC and EBU loudness scales.



In the mood

In this week's feature, Professor Rumsey talks emotion and finds out whether an audio signal might have the potential to give us the tingle factor.


Standardising loudness

For years there was no official standard to combat the problem of broadcast loudness. Now there are several and while, as Kevin Hilton reports, they are all similar in technology and intent, there are enough differences to cause uncertainty.


3D audio: reinventing the wheel

In this week's feature Professor Rumsey looks at the sudden rise in popularity of the weird and wonderful world of 3D audio and how binaural recording might finally have come of age.