How to game the Spotify Algorithm to have louder mixes on streaming (along with some thoughts on true peak)

hi friends

just sharing some insight on "true peak" - since that whole "no master" topic came around i've been experimenting with some older concepts i've learned over the years from friends who master/mix on a pro level.

one of these concepts was how Spotify "penalizes" mixes that are "too loud" or have true peaks.

here are some interesting findings.

A PROPOSED THEORY I'VE HEARD ONLINE:

"Peaks over 0 DB causes Spotify to "penalize" loudness - and per the streaming guidelines, you should NOT have peaks over 0 db or else you'll be turned down by the Spotify algorithm!"

This seems to be a myth - and I tested manually turning down all of the true peaks, and running it through RX 11's streaming preview. Assuming the streaming preview is true to Spotify (I used the Spotify Normal preview mode), it made zero effect on loudness on the final product.

I personally ignore true peak limiting...I think it sounds awful in most cases, and an old solution was to manually turn them down which is a major pain in the ass. It seems like a waste of time.

THEORY #2:

"Okay, what about mastering to -14 LUFS - anything louder is just ruining the audio! Spotify is going to lower it anyways! The streaming standards say we should master to -14 LUFs!"

This is also a load of shit as most of you know, and to be ignored. YES - the songs get lowered by the Spotify algo - but we're making music for more than Spotify (think, beatport). 99% of artists sending masters into DSPs are pushing pretty extreme loudness. Not to mention, if Spotify no longer becomes the standard, then what? You gonna live with -14 LUFS masters?

You COULD theoretically have a "streaming" master, but every platform's algorithm is different, and I doubt my distributor and labels would love to deal with 12 different master files for each annoying streaming platform. I can't imagine the look on a label's face delivering a master for Deezer separately...lol.

ONE THING I did find in my research though...quieter parts in the master file can actually PENALIZE you on Spotify.

Wait - WHAT?

Here's how to "cheat" Spotify's algorithm to make your masters louder than others.

This is actually a bit confusing, but rock with me. LUFS are a loudness measurement meant to measure loudness based on human hearing - not a meter/voltage or average of such.

One part of that measurement standard is that it doesn't measure parts below a certain volume - around -21db gets gated out. So quiet breakdowns below this gate threshold do not count toward the LUFs measurement. Read that again and digest - there's a gate at around -21 DB, so quiet sections are NOT measured in your loudness. This means only the loud parts will be measured, which results in a louder average.

Here's a visual guide to help explain.

Check this screen shot from RX11 with breakdowns that are "too quiet" to the LUFS standard:

Song example with breakdown underneath the gate, measuring at -8.67 LUFS.

The Spotify algorithm adjusts the gain based on the measured LUFS - and in this case, it's measuring only the loud parts of the song, because it deems the quieter parts as irrelevant. Your measurement is now -8.67 LUFs, so Spotify will penalize you - 5.33 LUFS, to reach -14 LUFS.

Now, if we raise the volume of the quieter breakdowns right above the gate, we measure in at -11.28 LUFS.

This is because the Spotify algorithm now measures out the entire song (including the quieter parts), which now gives us a lower LUFS reading, even though we didn't change anything in the louder parts, we actually turned the quieter parts up louder!

Wait what - a louder mix means LESS penalty by Spotify?! Yep. See below:

Now the LUFs reads at -11.28 LUFS, which is not a mistake - it averages the quiet parts too.

This means you'll only be turned down 2.72 LUFs by the Spotify algorithm - a BIG boost compared to the other version of the song - you'll gain 2.61 LUFs onto the older master with quieter breakdowns.

In short, the LUFS algo does not account quieter parts. The hack is to ensure you're always above the gate it uses, this way you can get a lower LUFS score, while actually maintaining the same loudness in the loud parts.

This of course can be turned off on Spotify in the Settings - but the average user is using normalization, which means your music will stand out.

Now - you don't need to know this shit to make great art. I don't care about ANY of this stuff when making music...buuuut I do love audio engineering and this is something I like to geek out about in my spare time and share with yall. I do test stuff on my own music occasionally when it comes to mastering concepts.

This is a relatively new finding for me and I'm not sure if I will change the way I master music because I don't care much about Spotify - but I want to constantly promote ideas in here that challenge the way we work/master/mix :)

Enjoy!