Levels in Mastering for the Various Streaming Platforms (and what happened to the K-system of metering)

Sage Audio and Mastering the Mix both wrote great articles on the levels required for the various streaming platforms so I thought I would share these as this is a common concern for music producers. I’ll also tell you how I am rationalizing my mastering.

First, the link to this great article by Sage Audio: https://www.sageaudio.com/blog/mastering/mastering-for-streaming-platform-loudness-and-normalization-explained.php.

Mastering the Mix has an equally excellent one here: https://www.masteringthemix.com/blogs/learn/76296773-mastering-audio-for-soundcloud-itunes-spotify-and-youtube

(I also don’t use the K-scale or K-system any more thanks to the inventor of it, Bob Katz himself. Scroll to the bottom to see why.)

Updated: 12-November-2021

Here is how I am currently mastering my music:

  1. I want to avoid the “loudness wars” where nobody seems to win and all music sounds bland.
  2. I use Distrokid for my distribution so I want to provide them with one file per song that will work everywhere they distribute my music (Spotify, iTunes, etc.)

To fulfill these criteria, I need to

  1. Limit peak levels in my audio to -1dBTP (TP=true peak).
  2. Have an integrated loudness of -17 to -14 LUFS, but absolutely no more than -14 LUFS.
    • Again, this is what most services want, and it keeps me out of the overloud range (see loudness wars).
    • Music too loud is fatiguing to the ears and often lacks dynamic range.
  3. Make dynamic music with a loudness range (LRA) of greater than 7 (the bigger the better for my type of music).
  4. Have no more than -8 sort term LUFS.
    • If it were more, this would mean my music is all running very high in loudness over the short term. This can be somewhat fatiguing to the ears, by my experience. Think of it as a lot of high volume points in music, with some lows in between. This can be tiring to the ears. Think also of someone who speaks too loudly at you and has a lot to say.
  5. Use a loudness meter with a graph instead of a regular VU or dB dial-type (or number-type) meter.
    • I want to see where my music is overall and not just a point in time. A graphical meter will show me that.
    • Fabfilter Pro-L is ideal for seeing the loudness features of a whole track over its playing time.
    • Master the Mix Levels is perfect for getting real-time loudness measures as well as a host of other useful items related. I am not fond of the interface though.
    • I like the interface of the Youlean loudness meter and am using it more often.
  6. Have good stereo correlation which simply means that I don’t want phase issues that cause cancellation of the left and right stereo channels when played in mono.
    • O.k., I don’t think most people listen in mono, but that is not the point. The point is that even if someone is listening in stereo, they still can perceive a drop in volume or, worse yet, a muddiness, when there are phase issues.
    • Phase issues are important, but the metering can be hard to understand. Phase correlation meters read out in values from -1 to 1. 0 to 1 is o.k., but under 0 in the negative territory, you start to have phase issues. So how far negative can you go? I find that -0.5 is perfectly o.k. for me if it is sporadic.
    • Negative stereo correlation of a more-constant nature is common if your bass frequencies are too wide. I like to take my low frequencies to mono. How low? I’ve experimented, but the catch-all for me is lower than 250 Hz, they’re mono’d.

Using Fabfilter Pro-L

Here’s how I do it using Fabfilter Pro-L

This is my Fabfilter Pro-L2 screen showing all of the settings I use. In the next images I will go into detail about each item shown on this. Also, by the way, this is showing a well-mastered track played from left to right. You can see it limits to -1 dBTP in only a few spots (two, for a maximum of -0.3 and -0.4 dB each), has a loudness range of 14.1, an integrated LUFS of -16.6, and a short-term LUFS of -11.2.

At the bottom of the main screen you can see I am doing 8x oversampling with true peak limiting.

On the left of the screen you can see the gain slider (vertical) set to about 7.5 in this example. I shoot for less than 3 dB of limiting and in this example you can see two limited events of 0.3 and 0.4 dB reductions. If I keep these limits less than 3 dB, I find I cannot hear the gain reductions so I think this is o.k. with my ears.

In this segment from the lower right, you can see I am using the Loudness metering, true peak (TP) is on (green), and my output is limited to -1.0 dBTP. I have my dither set to off, because I use Goodhertz’s Good Dither following the L2 as my last plugin in the chain of my master bus. My display setting (just left of Loudness in the image above) is set to infinite mode so I can see the whole track and identify portions needing attention.
In this right side portion of L2 you will see I have -11.2 measured as the Max S (maximum short term LUFS) at the top. I am using the -14(Strm) scale which puts -14 LUFS as the target (maximum in my case). I have the lower meter set to read the integrated LUFS value and it measured -16.6 for this particular track. With this set to integrated, you get the measure of loudness range, LRA, that measured 14.1 for this track.

Using Newfangled Audio’s Elevate

I have also used Elevate.

By using both the limiter gain and drive sliders, these combine to give roughly the loudness increase in my music. The adaptive gain, speed, and transient features automatically adapt the levels of each to make my music both punchy, loud, but not overly compressed. This is a main advantage of Elevate in my opinion. Gone is my endless tweaking of compression and transient shaping prior to the final output – Elevate does this automatically, with a few key inputs (read the manual). The saturation of the spectral clipper increases the perceived loudness while keeping the dynamics.

I keep my ceiling at -1.0 dB (not as in the picture above), as that has not changed as a good point to have for the streaming platforms.

Mono the Bass

I like my music to sound good in the low end, so prior to my limiter, I sum to mono all frequencies below about 120 Hz. I say “about” because sometimes it is lower or higher depending on what I’m going for in my music. Having that grounded bass is good though, not only because of phase cancellation but also because it just sounds more three-dimensional.

I use Goodhertz’s mid-side processor to achieve this. All of the Goodhertz plugins are top-notch in my opinion and I use them throughout my tracks regularly.

I use Goodhertz’s Mid Side and set all frequencies below “about” (see above) 120 Hz to mono (not as the image above shows).

You can also use Fabfilter’s Pro-Q to do the same (see below).

I have also used the Pro Q equalizer and taken the frequencies below a certain point (135 Hz in this image) to mono. This image here shows that I put a low cut filter affecting only the sides, effectively making mono (mid/center) everything below (left side of image). I don’t use both Pro Q and Mid Side, but choose one or the other, usually based on the content.

I’ve also used Mongoose to mono the bass.

Loudness Metering Using Youlean Loudness Meter

I use Youlean’s loudness meter exclusively now because it gives a good display that allows me to see where my music is lacking in width or loudness. You can see the interface below showing the loudness.

Another good tool to try is Loudness Penalty Analyzer at https://www.loudnesspenalty.com/. They have a plug-in now too so it is even easier to evaluate your tracks.

What about K-metering?

Bob Katz developed the K scale for metering and I used it for a long time – until I read his book, “Mastering Audio, the art and the science, third edition” where he said on page 261 that he no longer endorses it (see excerpt image below):

(“PLR” is “peak to loudness ratio,” by the way.)

Bob Katz is a master at mastering, and his advice is priceless. I’d highly recommend you check this article: https://tapeop.com/interviews/116/bob-katz-bonus/

Other Informative Sources