Loudness & Target levels in Audio Production
Updated: Jun 28, 2022
Loudness is a rabbit hole leading to a wonderland of complexity.
In this article, I will touch on the surface to try and make sense of the most important aspects of loudness with regards to audio production.
TLDR;
There are many ways of loudness metering, and they all have different uses.
Use metering tools to measure loudness at different stages of your production, and apply adequate interventions to limit or boost signals at the right place and time to create a smooth listening experience that is loud enough to keep up with industry standards.
Going too loud will cause distortion.
Raising the volume is easier than lowering it.
Mix on low volumes to hear the things clearer.
Removing frequencies that won’t be heard anyway might make a mix sound louder, even if it is technically not.
Target level standards vary between genres and platforms. LoudnessPenalty.com is a great website for finding out how close your audio gets to the standards of a given publisher (a negative "penalty" is better than a positive one, as platform normalization may cause clipping when raising the volume for you.
Target levels are important, but in the end the most important thing is that it genuinely sounds good.
Let’s dive in.
Why you should care about loudness
You don’t want listeners to have to constantly raise and lower the volume while listening, so you’ll want to keep your audio levels consistent.
In today’s landscape of fast-food audio consumption, the industry standard is to be as loud as possible.
Absolute loudness vs. Perceived loudness
As you may know, absolute loudness and perceived loudness are not the same thing.
Absolute loudness is the unbiased, straight up volume.
But human hearing is limited to more or less a certain range of frequencies, and is geared towards certain sounds, like other human voices, for example.
Perceived loudness is when we take these human factors into account.
Human hearing ranges from around 20 Hz to 20 000 Hz and is particularly sensitive to frequencies around 2 000 to 5 000 Hz. This means that, even if played at the exact same absolute volume, a sound played at 3000Hz will be perceived as louder than a sound at 300Hz.
Humans appreciate the loudness of tones more so than transients. For example, a blip will be perceived as quieter than a longer sound played at the same absolute volume.
Every sound and voice covers a unique range on the frequency spectrum, and that’s why it ultimately comes down to the audio engineer to make decisions regarding what will sound most appropriate to human ears... Unless your production is meant to be enjoyed by other species, of course.
Tip: remember to mix on low volumes to hear each element of the mix clearer. This way, it’s much easier to identify what needs to be louder/quieter.
By removing “unnecessary” frequencies (often low ones) from a track/mix, the overall sound will likely be perceived as clearer, and even louder. Sounds that are hard to hear still add to the absolute loudness of the mix, and plugins like compressors and limiters will be triggered by those frequencies, whether we like it or not. This can be avoided by reducing those frequencies in the first place.
Tip: remember to use low-cuts and high-cuts (AKA low-pass and high-pass) to remove unnecessary sound that don’t add anything to the overall sound.
Too loud = distortion
While tempting to be as loud as possible, there is such a thing as too loud. Exceeding 0dB puts us in the danger-zone of distortion.
But distortion is not always bad. Sometimes a little bit of distortion can help in making the sound be perceived as louder. More about that later on the topic of clipping.
Metering
Because loudness is measured and perceived differently, it’s important not to stare yourself blind on meters, but they are extremely helpful in giving you perspective of what you’re doing.
Most digital workstations (DAW’s) have stock metering-plugins, and they are usually adequate enough for most productions.
The following are the most common loudness metering techniques. These are the principles you’ll need to understand to be able to master and export a production to fit industry standards.
VU (Volume Unit)
VU meters usually comes in the shape of a pin jumping across a panel, or a bar filling up and down, and measures the loudness of sound being played in a given moment. Very handy in order to quickly visualize changes in volume in an accessible way. The accuracy of VU meters are usually pretty close to human interpretation of loudness, including some depreciation of transients, but unlike human ears, VU meters appreciate all frequencies equally.
Peak
Measures the absolute volume peaks. Peak metering takes into account the instantaneous peaks that human hearing can’t, including all frequencies equally.
True Peak
Displays the volume peaks of audio when converted to analog, with human appreciation of frequencies taken into account.
RMS (Root-Mean-Square)
The integrated value of an entire waveform, or, the average loudness of a sequence, which represents perceived loudness more accurately than peak metering, since it’s not instantaneous.
Be careful not to measure sequences that are too large using RMS. A sequence, like a whole song, or podcast episode, may contain significant dynamic range with deliberately louder and quieter parts for good effect, and averaging all that would be pointless. Instead, measure the loudest section of the sequence for more useful insights.
LUFS (Loudness Unit Full Scale)
A measuring unit close to perceived loudness. LUFS is the universal standard measuring unit when it comes to perceived loudness, and correlates well with decibels, making it easy to make changes that translate into the results you want.
LRA (Loudness Range)
Measures dynamic range using loudness units (LU). Dynamic range is the distance between loud and quiet. It’s application is entirely subjective in music, but in audiobooks, podcasts and broadcasting there are limits to consider. Loudness units are similar to RMS, as they measure loudness over a short period of time, but they are also adjusted to the human range of hearing frequencies to better represent perceived loudness.
It is important to understand that these ways of metering have their own time and place. Those who wish to master audio must learn when to use which.
Target levels
Target levels vary between genres and publishing platforms - and even the times.
Here are some solid reference points in general:
Podcasts
Average: -16LUFS
Peak: not exceeding -9LUFS
Audiobooks (ACX standard)
Average: -23 to -18 RMS
Peak: not exceeding -3dB
Music
Check each individual publisher/platform, as industry standards are constantly changing.
Voice Over
For commercials, as loud as possible, no dynamic range (AKA smashed).
For other use, check each individual publisher/platform.
What about recording?
Generally speaking, while mixing it is easier to raise the volume, rather than lowering it. So don’t overthink it, just make the initial recording sound good, and worry about target output levels later.
Controlling loudness
Moving a fader is perhaps what first comes to mind, but since loudness is a lot more complicated than that, so is the way we control it.
I’ve already talked about frequencies and how they affect absolute- and perceived loudness respectively, so now we will focus more on the sheer volume.
Compression
Dynamic range is not the same as loudness, but does play a big part. Compressors help us reduce the distance between loud and quiet by automatically lowering volume exponentially as it gets louder.
The opposite of compression is upwards compression, which is the process of increasing dynamic range, rather than decreasing it.
Compression can be used at different stages in a mix, and deciding how much dynamic range to keep is a matter of creativity.
Multiband Compression / Dynamic EQ
While not exactly the same thing, multiband compression and dynamic EQ perform a similar service, namely splitting up the process of compression into different parts over frequency spectrum. This allows us to apply compression to a select range of frequencies.
The difference between the two is that the multiband compressor is a compressor first, with additional features for splitting up the frequency spectrum into so-called “bands”, while the dynamic EQ acts as a traditional equalizer first, with added features for precise compression. Similar, yet different.
Limiter
The process of limiting is the final step in solidifying even volume levels. This is achieved by catching peaks in volume that would go above 0dB, and bring them down right below 0dB again.
If the mix isn’t loud enough to ever exceed 0dB, use the input gain of the limiter to push it up until it does.
Sounds like compression, right? It’s very similar, but traditionally the limiter doesn’t change the volume exponentially, but rather brings down only the excessive loudness.
Many modern compressors actually have built in limiting as well, so the lines do get blurred sometimes.
In addition to traditional limiting, there is clipping. This kind of limiting introduces distortion when active, giving the excess loudness a bit of crisp at the top when bringing it down.
In conclusion
Loudness control is extremely important. It’s something the listener won’t think about if done well, like so many other aspects of audio production.
Understanding how frequencies and human appreciation of frequencies relate to loudness is vital to and appealing sound.
Being as loud as possible is often important, but in the end, it’s about making a pleasant listening experience where people can enjoy the ride seamlessly. And if you do it well enough, people will turn up the volume for all the right reasons!
Let us know if you found this article useful!