Kim Lajoie's blog

Mastering for loudness. Don’t do it. Or if you have to, try this…

by Kim Lajoie on May 5, 2014

While mixing is the process of making sure the sounds in a mix are clear and well- balanced, mastering is the process of making sure each song on a release is clear and well-balanced with the other songs on the release.

The tools available to a mastering engineer are similar to those used by a mixing engineer, but are often more subtle and precise. They have to be – they’re used for processing complex audio (the whole mix). Compressors design for mastering are usually much more gentle; the sound of extreme compression on the whole mix is almost always undesirable. Similarly, EQ design for mastering is usually a lot more precise; tonal changes to the whole mix usually affect many different individual sounds and can modify the mix balance in complex ways.

Part of the role of a mastering engineer is to make sure the final playback level of the mastered audio is appropriate for the style of music. Acoustic music like classical and folk tend to have a lower level than modern highly-produced music such as rock and dance. For music that can have a lower level, there is greater headroom for peaks; the audio can have a higher crest factor. On the other hand, music that requires a higher level mus have lower peaks and lower crest factor. This means more loudness.

A mastering engineer’s primary tool for increasing loudness is the limiter. Conceptually, this is similar to a compressor with extremely fast attack and high ratio. Limiters are often used toward the last stage in the processing chain to ensure that the final audio level never exceeds 0dBfs. In mastering, the limiter’s sole purpose is to reduce the audio’s crest factor while sounding as invisible as possible.

For a good mix of a good composition, the mastering engineer shouldn’t have to apply too much limiting. It certainly shouldn’t be audible.

We start to push the boundaries for audio that has a high crest factor or when the executive producer wants the final audio to be louder than the level normally accepted for the style of music.

For these types of situations, regular mastering limiters can be inadequate. While they’re usually designed to sound as invisible as possible, extreme loudness will require processing that is audible. For these type of situations, saturation – or even clipping – will be necessary. This often creates a harsh sound as transients are crushed (distorted). Some digital limiters can combine or blend clipping with limiting, to provide greater gain reduction than pure limiters with less harshness then pure clippers.

Because the mixdown contains all the sounds of the mix as a single stereo audio feed, any changes to the audio affect all the sounds that are playing at that time. For example, a spiky snare drum that is crushed in mastering will also result in all the other sounds that are playing at the same time to be crushed as well – whether they need it or not. This is why this kind of processing in mastering should be a last resort – it’s much better to address these kinds of problems earlier on: in the mix or during composition.

In some situations, multiband limiting is appropriate. This is a crude attempt to contain the audible effects of extreme limiting to a subset of the mix. Using multiband limiting, a spiky snare that requires more limiting than usual won’t result in the bass being simultaneously heavily limited. This approach can sometimes be necessary for addressing problems that would have otherwise been best fixed in the mix.


2 thoughts on “Mastering for loudness. Don’t do it. Or if you have to, try this…

  1. Dave King says:


    Following up to your post… When mastering, do you aim for a particular RMS level in your mixes? If so, what is it? Curious minds want to know… Thanks!

  2. Kim Lajoie says:

    Hi Dave,

    When I’m mastering, I reference against commercial tracks chosen specifically for the job. There’s no one-size-fits all. As I wrote in the post, different styles of music have different norms and different listener expectations. That goes for both dynamics (including overall level, dynamic range and crest factor) and tonal balance.


Leave a Reply

Your email address will not be published. Required fields are marked *