Audio compression vs. normalization and leveling

What´s the difference between compression, normalization and leveling? A walk-through of when to use what function, and where to find good applications for the task at hand.

Anders Eklöv
KeyPleezer Music Tech Blog
6 min readJan 20, 2018

--

Ever wondered why there are so many tools for treatment of volume and sound dynamics? Well this post aims to shine a light on why and what to use to get the proper job done.

Volume Differences & Treatment

The core concept of audio dynamics treatment is to level out the differences in amplitude in the recording at hand. This is especially important today as we have an audio landscape in media where sound plays a big part in how we perceive information and often the quality of a production, whether it´s a video clip, a movie / series or a song on the ether. The 3 biggest conceptual methods for achieving a sound that is even and loud are called audio compression/limiting, audio leveling and audio normalization.

Audio compression (as in using an audio compressor, not lossy compression, like MP3) is not to be confused with normalization or leveling. While they are aimed in the same direction, which is to even out the differences in volume and amplitude within an audio recording, they go a bout it in different ways or handle different aspects of the process. Where compression/limiting is mostly aimed at handling the immediate dynamics and peaks, leveling usually does the job of changing the overall parts inside of an audio file to better match each other, producing a whole file that is better internally, as a block/clip. Audio normalization is aimed at raising or lowering the entire file/recording to a certain level. This is either from the Peak Level or the RMS level of the file.

Audio compression & compressors / limiters

So, audio compressors use the peaks or the immediate RMS levels (basically the general perceived volume of sound peaks) to stop sound from getting too loud. The compressor usually stops it in more fine ways, like with the option to set more parameters such as attack, release and knee functions, whereas the limiter does the job of stopping all sound at a given level, usually at 0 dBFS. This is used to stop the sound of a voice recording, a guitar or a whole mix from reaching a level where the sound would create digital clipping, which sounds horrible and flattens the sound peaks, leaving little detail.

Learn more about audio compression and limiting for music, podcasting and video here!

Audio Peak and Loudness Normalization

In contrast to compression, the normalization process does not do any dynamics processing of any sort, not as a general rule. It only raises the volume of the entire sound clip, or recording, to a specified level. The default of most music mixing, recording or mastering applications would be 0 dBFS.

Nowadays, with broadcast media and online streaming, a necessity to normalize sound volume based on a loudness standard, instead of the peak or even RMS levels of recordings or mixes, has given way to new Loudness Normalization standards and recommendations. There are numerous loudness standards for broadcast media and it depends on the geographical market in question and the material at hand whether to adhere to one standard or another. The concept of Normalization can also be split into Loudness Normalization and Peak Normalization, where the first would most avidly be used for broadcasting and video, the latter for music production.

Another important usage of Loudness normalization is while mastering music, where it´s used heavily to even out the perceived volume levels between different masters, so volume does not cloud the perception of the listener. Normally louder sounds sound better to most people.

Another use for it is to treat a large number of MP3 files, to even out the difference in volume throughout the library. It´s a common function included in most format conversion tools that convert wav/aif/caf files into compressed formats for web distribution or streaming, such as MP3, M4A/AAC and OGG.

Volume normalize plugin in Sony Sound Forge for Mac.

Wavelab´s Meta Normalizer can handle normalization in batch as well, which means to handle multiple files at once. The settings allow for skipping certain amounts of peaks, which can help make the loudness more accurate, especially with videos with explosive sounds that only occur rarely. Wavelab has many good tools for loudness manipulation.

Magix Sound Forge (formerly Sony Sound Forge) has a good normalizer plugin as well. Here showing how choices can be made to either peak or RMS. The “scan settings” allow for bypassing of extreme peaks. The setting “Use equal loudness contour” allows for compensation according to the psycho acoustic phenomenon known as equal loudness curve, or Fletcher-Munson Curve. This refers to the fact that we humans hear differently depending on the frequency. We tend to hear bass and treble less than the upper middle range frequencies. This is why we tend to turn up the volume when we want to hear the bass and presence.

Sound Leveling — the process of leveling out differences in phrases and internal audio parts

While the normalization handles the file in it´s entirety, the leveling process is aimed at evening out the different phrases and broader parts within an audio capture or recording first. Either your voice-over recording or podcast has too large differences between sentences or the people in the room is at different distances from the microphone. You would certainly use a compressor for the entire mix to even out the peaks and get the whole recording in check or ready for broadcast, but it would be a good idea to use a leveler to even out the parts first. Many levelers, like the one in iZotope RX, have this ability.

iZotope RX6 Leveler does a good and custom job at leveling audio phrases. Very useful for many sources.

The leveling process is essentially the same process that your would apply when using clip-volume or automation directly on your recording track in your DAW (Digital Audio Workstation). But whether you use a multi-track recording software like Logic Pro or Pro Tools or any single-track audio editor, the process is the same. It´s simply the overall adjustment of the volume level over an entire phrase, set of phases or even the entire voice track in a podcast or audio book. Using a leveler can save you a huge amount of time.

KeyPleezer is a producer of virtual studio instruments and Soundware for Samplers and DAWs on PC & Mac.

We have a free piano plugin/soundware awaiting your music production studio.

The LivingRoom Upright Piano — free edition, for Kontakt sampler & EXS24 (and LPX 10.5+ Sampler). Also available with dedicated patches and installer for Apple Logic Pro X, GarageBand and MainStage!

Get it now for free on KeyPleezer.com!

--

--

Anders Eklöv
KeyPleezer Music Tech Blog

I´m a music production enthusiast, mixing and mastering engineer, creator of KeyPleezer and an all-round tech tinkerer with a big interest for music technology!