I found a really interesting video in my usual internettings this week about semi-recent developments of loudness in audio. The original post is a couple of years old now but I think the information is still relevant and likely still a mystery to many of us, it was certainly news to me.
I am going to pick out the salient points in the video to help us break down the jargon a bit, and I have included the full video at the bottom of the post for any that want to hear it from Ian the Loudness Warlock, directly.
When mixing and mastering there are quite a few ways to measure loudness in the sounds or music that we are writing. In recent times, the main two methods for measuring the peaks and troughs are dBFS (decibels Full-Scale), also known as Peak, which measures the moment peaks of a waveform, with a maximum of digital zero; and RMS (root mean squared) – that means average to us mortals. RMS gives us the average reading of the loudness over a longer period of time.
Another important value is DR, or Dynamic Range – this is the difference between the quietest sound and the loudest sound. I’ll touch on this a bit more later.
Over the last few years a new standardised metering system has surfaced, called LUFS (Loudness Units Full-Scale). To keep it simple, LUFS values will basically be the same reading as RMS values, as they take an average loudness measurement. Also good to know is, if you drop your fader by 1dB, your LU meter will drop about 1LU as well. This makes it easily translatable from one scale to the other.
LUFS is actually smarter than RMS, as it has two timescales that it measures across. The first is the integrated loudness level. This is similar to the RMS measurement, which has a longer timeframe and is slower to react to changes in the loudness – this gives a great reading of the average loudness of a whole track for example. Ian says that any track that has an integrated LUFS reading of -11/12LU is pretty darn good.
The second timescale is the short-term loudness, which as you would imagine has a shorter timeframe that it measures across; it therefore reacts more quickly to changes in loudness. This is useful in practical terms to measure the difference in loudness between two song sections – for example it would show you how much of a difference there is between a second chorus and a middle eight – useful info indeed when mixing. Ian says that short-term LUFS reading of -8/9LU is spot on.
Physiologically, the peak values are not that important to us. Our hearing is not sensitive enough to notice the moment to moment loudness changes enough for it to affect our experience, as long as the peak does not exceed digital zero and cause nasty digital distortion. In fact, those momentary changes in the loudness are usually what cause us to enjoy the music more. The RMS values, and now the LUFS values, particularly the integrated LUFS are much more useful.
Away from mixing and mastering slightly and towards broadcasting.
LUFS have widely been accepted in the US and the UK as the standard broadcasting scale.
-23 LUFS integrated is now the standard in the US for TV, and I have found a white paper published by the BBC from 2011 which shows that they too are committed to that level:
Basically, what this means is that any track which exceeds an integrated LU (remember that means average) reading of -23LU will be decreased in loudness when it is broadcast; and any track that is below that level will be increased, so that integrated LUFS are the same. Crucially, dynamic range (that is the difference between the loudest sound and the quietest sound is kept in tack. For tracks that are overly compressed, and suffer from the fallout of the loudness war, the outcome of this would be that the entire track would be turned down from around about the -7LU mark to that -23LU value. Because of the overcompression, the dynamic range has been lost, and so at -23LU this track would be far more boring. This is great news for genres such as classical music, where the dynamic range is much larger, because listeners will now be able to turn up their speakers, without the worry of the next song blowing their ear drums! See picture below from Loudness Alliance’s white paper from July 2012:
In Summary
- RMS and integrated LUFS are pretty much the same in terms of reading values
- LUFS give two timeframes that you must be aware of, integrated and short-term.
- Peak readings are great for making sure your recordings don’t clip, but don’t give much information about the loudness of a track.
- -23LU is the magic number when broadcasting, and this is a great development for clawing back some of that headroom, and dynamic range.