ONLINE ACADEMY FOR AUDIO ENGINEERING & MUSIC PRODUCTION

Mastering for Spotify, Apple Music & More – the Loudness of the Pros

In this article we want to bring clarity to the topics of mastering, loudness and streaming. We will also show you what requirements the streaming providers have for your master, what happens to the loudness of your song on the platforms, and what loudness levels the pros actually go for.

To explain the basics of this topic, let’s take a brief look at the concept of loudness.

What does loudness mean?

We are in the digital age, and audio signals are represented by a limited number of ones and zeros. This results in a maximum representable value, also known as full scale (0 dBFS). This determines how loud we can make a song without clipping it or affecting its dynamics.

But why do different songs sound louder or quieter when normalized to the same full level? This is due to the subjective “loudness” that results from the average level. If the quietest levels of song A are higher than the quietest levels of song B – in other words, if song A is less dynamic, song A will have a higher average level and therefore a higher loudness. This loudness can be measured (e.g. with the IQ-Series Limiter) and expressed in numerical values called LUFS (Loudness Units relative to Full Scale). When we want to evaluate the loudness of a song, we usually talk about “LUFS Integrated”. This value indicates the loudness averaged over the entire song.

Left: dynamic audio / right: little dynamics

Compressors or limiters are used to process these dynamics and influence the loudness or LUFS value. The mastering engineer is responsible for limiting the mix and determining the loudness at which the final product will be heard on the various streaming platforms or physical media.

With our current special offer, you will get the HOFA IQ-Series Limiter V2 (and all other HOFA plugins) as a gift when you register for the AUDIO DIPLOMA online course.

In practice, you want a master that is as loud as possible, because a louder track is generally preferred by the consumer in a direct comparison – the song attracts more attention and is often perceived as more powerful.

However, a high loudness level can have disadvantages as well, and at a certain point strong limiting can cause unpleasant side effects in the form of audible artifacts.

So much for an introduction to loudness – you can learn more in the mastering topic in the AUDIO DIPLOMA. Now let’s bust a few myths and show you how it’s really done.

What are the specifications of the streaming platforms?

If the loudness levels of different songs on the streaming platforms were too different, the end user would have to constantly adjust the volume control to hear all the tracks at the same volume. To prevent this, streaming providers require a certain loudness (measured in LUFS) for uploaded tracks.

-9 LUFSSoundcloud
-11 LUFSSpotify (“Loud”)
-14 LUFSSpotify (“Normal”), Amazon Music, Tidal, YouTube,
-15 LUFSDeezer
-16 LUFSApple Music
-19 LUFSSpotify (“Quiet”)

This raises the question of what happens if these guidelines are not followed and the master has a different loudness.

Loudness normalisation

Depending on whether the measured loudness is above or below the platform’s loudness target, the master is simply made quieter or louder. This is called loudness normalisation. On Spotify it looks like this:

  • Louder than -14 LUFS: The song is levelled down so that it has a loudness of -14 LUFS.
  • Quieter than -14 LUFS: The song is brought as close as possible to -14 LUFS with a level boost, provided that the true peak is a maximum of -1 dBFS.

The values specified by streaming portals can therefore be exceeded without any loss of sound quality due to the normalisation process. This brings us to what many will find a shocking truth: No, you don’t have to master to -14 LUFS! But what use is that if our song is going to be reduced in level anyway, so that it will be played back at -14 LUFS like all the other songs?

Why should you master louder than -14 LUFS?

-14 LUFS ≠ -14 LUFS

Even after loudness normalisation, the songs may sound differently loud because the LUFS value does not provide a 100% reliable statement about the subjectively perceived loudness in reality. This is due to a number of factors:

1) The loudness during the song

For songs with a high program dynamic range or high loudness range (LRA), the loudest parts (e.g. the chorus) will be louder after loudness normalisation than the loudest parts of songs with a lower program dynamic range.

When measuring loudness – according to the ITU 1770 standard used by streaming platforms – two “gates” are used. These define a threshold for the levels in the song to be included in the loudness measurement and ensure that downward “outliers” are ignored.
The first gate is set at -70 dBFS, excluding silence.
The threshold of the second gate is 10 dB below the LUFS value measured up to that point.
The final integrated LUFS value therefore also excludes everything below this second, individually set threshold. The following trick can be used to make the platforms sound louder: make the parts of the song that are below the threshold of the second gate a little louder so that they are above the threshold. As a result, these quieter parts are now included in the measurement and reduce the measured average loudness (LUFS integrated). On streaming platforms, this means that the level reduction during loudness normalisation is less pronounced. As a result, your song will play louder. This trick is illustrated in the graphics below:

Before (left): 50% measured, -8.67 LUFS (played back 5.33 dB quieter on the platforms)
After (right): 96% measured, -11.28 LUFS (played back 2.72 dB quieter on the platforms)

2) The frequency distribution

Unlike a VU meter, an LUFS measurement includes frequency weighting. The well-known Fletcher-Munson curves (equal loudness curves) would be ideal here, but “only” the so-called K-weighting is used. The K filter consists of a high pass at 80 Hz and a 4 dB high shelf at 2 kHz. Two masters measured at the same volume can therefore appear to be louder due to a certain frequency distribution.

3) The Crest factor

This is a measure of the relationship between the transients (peak) and the current average level (RMS). A lower crest factor can make a song sound denser and louder – even if they have the same Integrated LUFS value. However, transients can suffer if the crest factor is too low.

Avoid streaming limiters

In addition to the default setting of “Normal” (-14 LUFS), Spotify even allows you to set the loudness normalisation to “Loud” (-11 LUFS). In this case, tracks that are quieter than -11 LUFS are brought up to this level using a limiter. However, this type of sonic intervention in the master should be avoided, as the mastering engineer should remain the final sound editor. If the master is louder than -11 LUFS, it will not be affected by a limiter.

Normalisation cannot always be relied on

In addition, loudness normalisation is not always necessarily active on the platforms. On the one hand, you can switch it off manually in the settings, and on the other hand, there are a few exceptions (e.g. game consoles) where normalisation is not implemented. In this case, your master would also have to have a higher volume in order to keep up with the volume.

Enough theory. What does it look like in reality – what are the values of today’s successful productions? Here is a little insight:

Statistics/references

While the average loudness in the 70s was still around -20 LUFS, the most popular songs today have values of around -8 LUFS. Let’s take a closer look at the “modern era”. To do this, let’s look at the Integrated LUFS values of a total of 45 songs that were released in 2024 and were largely represented in the Billboard Hot 100 or the German singles charts.

In this chart, we can see that most songs have a loudness of -9 LUFS to -7 LUFS. The extreme values in our analysis are -11.2 LUFS (Drake – Family Matters) and -4.6 LUFS (Knocked Loose – Suffocate).

A look at the distribution by genre is also interesting. However, it should be added that genres that are only represented by a few songs are not equally meaningful. With a total of 18 songs, the “pop” genre is the one that is most likely to provide a statement. Here it can be seen that even songs from the same musical style that were released in the same year and have a similar level of success have very different values – from -10.7 LUFS (Tom Odell – Black Friday) to -6.0 LUFS (Nemo – The Code).

Another interesting value would be “Short Term LUFS”. In contrast to the “Integrated LUFS”, this figure makes a statement about the loudness within a short time window, which is set at 3 seconds by EBU R128. For the songs from this statistic, the range of the measured maximum Short Term LUFS values is -8 LUFS to -3.5 LUFS.

Conclusion

In the end, the song should just sound good – sometimes a song needs a very dense sound and sometimes a song needs more dynamics. So you can “fall off the horse on both sides”: It doesn’t make sense to “squeeze” a dynamic song that sounds good at -12 LUFS to -8 LUFS with artifacts just because the current No. 1 hit is mastered to -8 LUFS, nor is it reasonable to master a song that needs a super dense sound to -14 LUFS because that’s the “required value” from Spotify.

Nevertheless, it makes sense to take a look at the Integrated LUFS from time to time and keep the approximate value range of today’s successful productions in mind. With values between -12 LUFS and -6 LUFS, you are generally in the green zone as far as the listening habits of modern, successful productions are concerned. Nonetheless, the golden rule is that, as a beginner, you should hold back on over-processing if in doubt.
If you want to gain more confidence in your productions, our audio engineering course is just right for you!

Author

Marvin Müller
Marvin Müller
Marvin graduated with a Diploma in Music Production and Sound Engineering as well as a degree in Electrical Engineering and Information Technology. His hands-on experience in the studio and theoretical knowledge of technology puts him in an excellent position as an audio engineer and tutor. At HOFA-College, his focus is on course development and mix analysis.

Leave a Reply

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