On The Subject Of Loudness

It’s been a long time since I’ve written anything here.  Not because I had nothing interesting to say, it’s just that it’s been incredibly busy.  Every year I seem to say that the last year has been the busiest ever, and then the next one is even busier.  Anyway, busy is good!

Every year, Sony Worldwide Studios runs an internal tech conference alongside the Game Developers Conference in San Francisco.  I spoke there a couple of years ago on the problems we as developers, publishers and platform holders had with loudness on our platforms.

At the same conference, the subject of loudness was raised by another couple of speakers, and we decided there and then to form a group to look at the issues.

So in March 2011, the Sony Worldwide Studios Audio Standards Working Group (ASWG) was formed, consisting of Gene Semel & Marc Senasac (SCEA), Keiichi Kitahara (SCEJ) and myself representing SCEE.  The first task we set ourselves was to address our loudness problem.

In the last couple of years we’ve made considerable progress.  Firstly, we studied the state of loudness in current media, measuring the loudness of over 120 games, TV shows, films and adverts.  In November 2012 released our first paper entitled ‘Average loudness and peak levels of audio content on Sony Computer Entertainment platforms’, based on the ITU-R BS.1770 and EBU R128 specs.

Measuring the loudness of interactive content is an inexact science, and we’ve been working with closely with our own First Party Quality Assurance departments on how to effectively measure the loudness of the titles we develop and publish.

Internally, our recommendations have been adopted by all first party studios, worldwide.  They’ve also been received very well by the wider industry, and we’re currently in discussions with a number of industry bodies about wider adoption.

So, I’m doing a number of talks over the next couple of months about the work we’ve been doing, and how we measure loudness in interactive media, both in production and at QA.

On Thursday 7^th Feb 2013, I’m speaking at the Audio Engineering Society 49th International Conference on Audio for Games in London with an imaginatively titled talk, ‘Measuring Loudness in Interactive Entertainment’.  I’m also on a panel the same day entitled 'Theoretical,Technical & Practical Frameworks for Interactive Mixing: A Moderated PanelDiscussion', with Tom Colvin, Jon Olive, Xavier Buffoni, Stephan Schütze and moderated by John Broomhall.

At the end of March, I’m at the 2013 Game Developers Conference at the Moscone Centre, San Francisco hosting the Audio Boot Camp with Scott Selfon from Microsoft.  I’m also on a couple of panels; the first with Mark Yeend (Head of Audio, Microsoft), Gordon Durity (Audio Director, EA) and Adam Boyd (Audio Director, Activision) about loudness standards in the games industry, where we’re at and the progress we’ve made so far. The second is entitled 'How to Mix a Video Game - And Not Die Trying' with Juan Peralta (Skywalker Sound) and Rob Bridgett.

Once this two months of madness is done I'll post up a transcript of my loudness talk here.

What's the difference between LKFS and LUFS?

As part of my work, I've been reading a lot about audio loudness standards recently.

There seems to be a bit of confusion between LKFS and LUFS, so I thought I'd explain the difference between these two measurements.

So, the short answer to the question, what's the difference between LKFS and LUFS is.....nothing.  Not any more anyway.  They used to be different but now they are effectively the same thing.

Here's the long answer:

About 4 or 5 years ago, the ITU (International Telecommunications Union) released a paper, 'ITU-R BS.1770 - Algorithms to measure audio programme loudness and true-peak audio level.'  It described a way to measure the average loudness of a piece of audio and gave you a value for the audio as a whole.  The units it used were LKFS, which stands for Loudness, K-Weighted, referenced to digital Full Scale, averaged out over the duration of the piece.

The EBU (European Broadcasting Union), took this and ran with it.  However, over time they discovered a problem.  It was too easy for long quiet sections of audio to bring down the average measurement.  So they released a paper, 'EBU R128 - Loudness normalisation and permitted maximum level of audio signals', containing a few modifications.

They introduced some new units, namely LUFS (Loudness Units, relative to digital Full Scale), LU (Loudness Units), and LRA (Loudness Range, a measurement of the dynamic range, measured in LU).  The LRA measurement was actually developed by TC Electronics, who are part of the EBU PLOUD group.

The big change was the introduction of a gate that would stop data being taken into account if the level dropped below a certain threshold.  Because consumers judge the loudness of any particular piece of audio based on the loud parts, what the EBU called the foreground loudness, this gate would ignore any data that was 8 LU below the ungated measurment, so quiet sections would not skew the overall measurement.

So, at that point, there was a difference between LKFS and LUFS.  LKFS was the ungated measurement, and LUFS was the measurement that included the effects of the gate.  Generally, on your average TV programme (over a period of say 20-30 minutes), the difference between the LKFS and LUFS measurements is usually between 1-2 LU.

However, in March 2011, the ITU updated their paper with ITU-R BS.1770-2, incorporating the changes the EBU recommended in it's paper R128, namely the gate.  Both the ITU and the EBU agreed that they would both change the threshold of the gate to -10 LU.

So, as of March 2011, LKFS = LUFS.  The EBU were hoping that the ITU would drop LKFS and adopt the EBU's LUFS units, as Florian Camerer, the chair of the EBU's PLOUD group told me that the ITU spec doesn't follow standard naming conventions for units.  However, they didn't, so we're stuck with both of them.

Personally I find it a lot easier to speak of LUFS, as it's easier to describe the differences in material as being 3 LU, as opposed to 3 L_lowercase K_.  But hey!  Anything is better than nothing!