Business Interviews Labels & Publishers

Q&A: Bob Katz

By | Published on Thursday 7 November 2013

Bob Katz

Speaking at The Audio Engineering Society Convention in New York last month, mastering engineer and President of Digital Domain Bob Katz made a bold claim: That the so called ‘loudness war’ will be over by mid-2014, and that it will be iTunes Radio that makes this happen.

‘Loudness war’ is the term given to the trend for recordings to be mastered increasingly loudly over time, as artists (or at least their associates) strive to make their records louder than anyone else’s and thus make them stand out from the crowd. The problem is that, the louder the music gets, the less subtlety and definition there can be in the recordings.

Katz, who has worked in audio engineering since the early 70s, as well as teaching and penning several books on the subject, told AES that through testing he had observed something interesting about the way Apple’s iTunes Radio service processed the music it plays.

iTunes’ Sound Check algorithm normalises all songs to the same level, so that volume doesn’t jump up and down as the player cycles through them. This brings the volume of quiet songs up and pushes louder songs down. And perhaps more importantly, this process can’t be turned off.

This, says Katz, is the key to ending the loudness war. To find out why, CMU Editor Andy Malt caught up with Katz to talk audio mastering.

AM: First, could you just briefly explain what you mean by ‘the loudness war’? What does it mean for music fans? Has this kind of mastering now become the norm, would you say?
BK: There has always been a loudness war, starting from the very first acoustic recordings of the 20th Century, because artists want to be heard above the fray. The level of vinyl LPs went up about 4dB to 5dB throughout their lifetime. This rise was limited by mechanical constraints though, because the needle would jump out of the groove, a sure discouragement to being too loud! So the maximum difference between the loudest and softest LP was no more than about 6dB, which is easily tolerable by music fans.

But the digital world has changed everything. The processing tools that are available digitally have allowed a loudness race that has created a difference between the current loudest and softest digital recordings of almost 20dB! This is the difference between a shout and a whisper. Consumers put on an oldie CD followed by a recent popular release and they could easily blow their ears or their loudspeakers. The digital processing required to achieve these extreme levels also creates a great deal of distortion.

Of course there is nothing wrong with artists making a record that has their sonic vision and however much distortion they want to create, but, the more processing that’s applied, the louder the record becomes. Because of this, and the desire to be loudest, the distorted approach to sound has become the norm in record production and mastering; and very few artists are bold enough to turn down, their natural insecurities usually outweigh their sonic and artistic desires. So one artist’s freedom of expression paints another artist into a corner and he or she often finds it necessary to process a record beyond their own sonic desires in order to make it compete loudness-wise.

The whole situation has become an obstacle to artistic integrity and freedom, and a disservice to all consumers. The only ones who may benefit are the artists with the loudest record, though even then only briefly, ie during the interval between when the consumer puts on that record and turns down his own volume control! Ultimately the artist has never had control over how loud his or her record will be heard, never did, never will.

AM: How does Apple’s Sound Check algorithm work?
BK: Radio and television sound has always been processed to equalize volume. But the method which was applied was to squash and compress the already-compressed recordings so that what comes out of the radio sounds particularly distorted and small. The combination of traditional radio processing and already-distorted sources is as lo-fi as you can get. That’s why the method which is being applied by Apple is particularly attractive.

Apple’s method is to set a “target” level for all their playback. Loud songs are brought down to this target, and soft songs are brought up to the target (soft songs can only be brought up to the point where their peak levels would not overload). No other processing is performed. At first I measured what I thought was some peak limiting, but this was due to some bitrate issues that I did not take into account. No other processing is performed. This is a big development, sound quality of the original recording is maintained, which preserves more of the artistic creativity of the original artist than ever before in the history of broadcasting.

As an aside, radio broadcasting in Europe and other parts of the world is quickly adapting to the same approach as Apple, soon leaving traditional radio in the US the only broadcast medium that will add its own distortion. I hope that traditional radio stations will adapt to the European approach, or US radio will rapidly become the dinosaur of the industry. iTunes Radio demonstrably sounds better than most US broadcast stations today.

AM: How does Apple differ from other services, and why is it important?
BK: Of course Apple was not the first to loudness normalise. Pandora has this as an option. But Pandora’s target is too high to permit good sound quality or loudness normalisation.

And, most importantly, Apple has made Sound Check a default in iTunes Radio. It cannot be turned off. This is a big development. It levels the playing field for all musicians. And we engineers hope that Apple will soon turn it on in regular iTunes as a default too, which will hasten the end of the loudness race. Apple’s influence will hopefully lead the way as Google, Amazon and other streaming providers enter the race. Let’s hope that the different Internet radio stations don’t enter into a loudness race of their own!

There’s more to this discussion, of course, but these are the basics. I cover a lot of this topic in more detail in my book, ‘iTunes Music’. Or, join in the forum at and express your feelings and criticisms.

AM: How will iTunes Radio and its use of Sound Check persuade artists and their labels to master differently for digital?
BK: I should say, Apple isn’t asking that tracks be mastered differently, and the success of their service isn’t likely linked to mastering. And songs which have been mastered in any flavour play well on iTunes Radio.

But songs which have been mastered more conservatively will probably sound even better, a little clearer and a little louder on iTunes Radio. Note, this has been the case even on regular radio – the best-sounding station on the dial is usually the oldies station!

However, what will happen, all by itself, is that musicians and producers will discover that on iTunes Radio (like other normalised media) a more conservative approach to levels will sound better. And it will happen because we will hear the improvement. Squashed songs won’t sound as good next to their competition.

Over time, recordings will return to sounding more open, dynamics will return to music – simply because we can hear the improvement. Many engineers, artists and producers have been yearning for years to have the yoke of over-compression removed from their necks. It will take a while, but we can now see the light at the end of the tunnel.

AM: What effect do you think this will have on mastering in general?
BK: Many engineers and producers are under the misconception that if they produce loud masters it will make them “radio ready” because they think that will make them sound louder on the radio, but that is a complete misconception.

Unfortunately, many engineers did not buy the argument by the evidence of traditional radio, but they will be convinced of the futility of making loud masters by iTunes Radio (and other loudness-normalised streaming). It’s simple: Now we can instantly compare the same song on the same device and loudspeakers that we use to play radio and to play file-based media. The evidence will be obvious – if we push it up, the radio will pull it down!

For a while there will be inertia in the audio community to continue to produce loud CDs. But it makes no sense to do it: Radio, the internet and clubs are increasingly the main media by which consumers get exposed to music. With radio and internet, music will be increasingly loudness normalised. In clubs, the DJ has control over the volume. So there’s no point. The more media that are loudness-normalised, the more we will take it for granted and eventually the loudness race will be a thing of history.