Master your metering
Want to get up to speed with today’s essential loudness measurements? You’re in luck – here’s our handy guide to VU, LUFS, K-Metering and more
VU and RMS
Standardised in 1942, the ‘Volume Units’ (VU) meter was deliberately designed to be slow and sluggish. The physical mass of the needle made it too slow to react to very short transients, so the reading tends to average out signal levels. Peak levels are almost irrelevant when it comes to perceived loudness – a long, sustained sound will always seem louder than a short transient at the same peak level, so you need a way to average out levels.
RMS (Root Mean Squared) meters take a short window of the signal, and measure ‘power’ in a way that correlates much better with perceived loudness than VU metering. The choice of window size makes a big difference, though: averaging over a longer period will give slower readings, and vice versa.
Mastering engineer Bob Katz proposed standard levels for the audio industry a few years back in the form of his K-Scale metering. This displays RMS levels on the same scale as peak levels, with three target RMS levels to aim for (K-20, K-14 or K-12), each providing higher absolute levels, with progressively less peak headroom. Katz’s K-14 and K-12 scales are still very useful for pop and rock mastering, but RMS has come to be superseded by a more sophisticated approach.
A new way
Around 2010, the broadcast industry got together and created an international standard method of measuring loudness (ITU-R BS.1770), which is now used to define European Broadcast Union recommendations (EBU R128). BS.1770 standards define a new unit of measurement called the Loudness Units (LU), usually displayed as LUFS (Loudness Units relative to Full Scale) or, occasionally, LKFS (Loudness, K-weighted, relative to Full Scale). One loudness unit is equal to one decibel, which makes life easier – with a well balanced mix, the LUFS readings will be very similar to RMS levels. However, RMS tends to read too high for bassheavy signals, which the new standard fixes with the use of ‘K-weighting’. It also includes a sophisticated gate mechanism, so that pauses don’t count towards the reading.
No BS
A new breed of loudness metering plugin is now available, based on the BS.1770 standards, including Signum Audio Bute Loudness Analyser and FabFilter Pro-L 2. They typically display three different readings. The ‘momentary’ reading is averaged over a relatively short 400ms window, and reacts quickly to short, loud events. The ‘short-term’ reading is averaged over three seconds, giving a much slower moving meter – this is perhaps the most useful when judging loudness for mastering. Finally, there’s the ‘Integrated’ measurement, which averages out levels over the entire duration, providing a single readout for a whole song, or even an entire album.
Your DAW or editor may let you select a clip or region and calculate integrated loudness offline. Otherwise, you’ll need to play your whole song into your loudness metering plugin to get an accurate integrated reading.
Integrated measurement can be misleading, as it gives a single value for the whole song. If a song has a quiet build-up, say, it will end up with a lower integrated loudness than a mix that’s full-on all the way through, which could lead you to make the ending too loud.
A better approach is to measure just the loudest section, or watch the short-term loudness reading on those parts. Pop and rock music usually sounds great with the loudest parts at around 10 to -12 LUFS.