SPOTIFY normalization, anyone has experience with that?

Svemir

Member
Hi guys,

so I'm having serious doubts about normalization in Spotify.

Everyone says, and it's also written on their website, that you should keep your master at around -14 LUFS integrated (on average), with true peak set to -1.

Now I have my master at around -13/-14 integrated LUFS, and checked on loudnesspenalty.com, it will normalize it just by -1 which is fine, even though for me it sound much more quiet then those reference tracks on Spotify.

The thing is, I downloaded from soulseek some reference tracks in FLAC format, and if I upload them in loudnesspenalty.com it will show me a normalization of -8!!! I'm speaking about big artists reference tracks.

Now I understand that the downloadable version of the reference artist tracks might be those mastered for CD or to be listened in local, which usually should be at around -6 LUFS integrated (very high volume) and maybe they have another lower volume version for Spotify, but if I go to their respective Spotify and listen to those same reference tracks on streaming, comparing it with the ones in FLAC downloaded which on loudnesspenalty.com were showing a normalization of -8, they have exactly the same volume!!

So here someone is taking a piss off me, Or Spotify normalizes tracks to -14 LUFS only if you don't have a premium account while if you have, you can even upload a -6 LUFS integrated track??
Otherwise why the hell the downloaded FLAC version of a reference track which shows a -8 normalization on loudnesspenalty.com, sounds the same volume as the one on streaming on Spotify?
You can make this test if you like.
 
Last edited:
There you go indeed, so Spotify won't normalize anymore, Bandcamp never did. Does it make sense still to master at -14 LUFS integrated?

 
Surely the point is to make all the tracks not require the listener to change the volume? I've often wondered if the algorithm they use simply identifies the tracks pushing the volume and this triggers the modification? My tracks are always quieter, and they don't seem to be very much different to what was uploaded?
 
I faced this problem while uploading my track first on Spotify because it won't let me even add it to my channel. I didn't have a premium account back then, so I thought the problem was particularly in my song. Good for me, I tried it again from my girlfriend's computer, where she had a premium, and everything worked. Spotify has a lot of such things where you're restricted without premium. Now I know most of them and even use them to my advantage. Also, I use a plan of gaining more audience at https://streamingfamous.com/buy-spotify-plays/. For only 25 dollars, I have 10k plays in my music genre. It's convenient for me because I'm not too fond of advertising but also want to be famous.
 
Spotify and the others are extremely good at spotting fake plays. Your music gets removed instantly - all of it, as in your entire catalogue, you have no right of complaint and any funds you thought you'd get get frozen. Paying for plays is now plain silly - far, far too risky!
 
If you grab tracks at random from Spotify, they are all different in absolute measurable maximum level unless they are the kind of track that when you look at it in the editor, is totally squashed the hell out of. Those tracks with a brick wall of sound seem to be a very similar level, but if you look at non-EDM style music that does in fact have a dynamic range, then these tracks seem to be much less messed with by whatever system Spotify (and others) employ. A bit like the classic radio Optimods when all of the protection kicks in. Lots of folk who put up music with dynamics never have a need to even worry too much about their levels - many have not even switched on metering that uses LUFS - they just use the normal meters and peak at -3dBFSor -6, or even -12 - whatever they fancy. 3dB is really not even worth worrying about. Drop a track of mine by 3dB and I doubt anybody would even notice. Maybe even 6dB would not really be picked up - it's all genre specific.
 
Here are the specs. Limiters for "soft" tracks that fall below spotify's -11 or -14 thresholds are set at -1.


So ... If you master the softest tracks to -11LUFS and don't ever exceed -9 lufs (or -7lufs if you want a brick wall of noise) for loud tracks, and all with a true peak of -1dbfs, the only thing Spotify will do is turn the volume down depending on user settings.

Loudness is easy.

Perceived loudness though ... That's a whole other ball of wax.
 
Last edited:
I understand your concerns about Spotify's normalization process. It can be a bit confusing at times. The loudness standards on Spotify are generally aimed at ensuring a consistent listening experience across different tracks.
Regarding the reference tracks you downloaded from Soulseek, it's possible that those versions were indeed mastered differently, possibly for CD or local playback, which could explain the -8 LUFS normalization you observed.
You can also explore services like https://songlifty.com/spotify-promotion/ to help boost your music's visibility and reach on the platform. Good luck with your endeavors!
 
Last edited:
Despite what Spotify recommend, most mastering engineers do not use the -14 integrated LUFS as a target level. It is a measurement that tells you at what level Spotify will play back the audio 'if' loudness normalization is activated. This is then used to make judgement on how the audio will sound when, or if, it is turned down (or in some cases turned up). The master will always be based on what sounds best regardless of the LUFS guidelines.

If the master track is at -8 iLUFS, the engineer can use the -14 iLUFS guideline to try a more dynamic version of the master. If the new version sounds better at say -10 iLUFS then it is a win win situation. He has created a more dynamic master and also got closer to the normalized playback level which means the track will not have to be turned down as much. If the master does not sound better at -10 iLUFS then it would not be used.

There are other factors to consider as well.
The Spotify web player and 3rd party devices like TV's, do not have any loudness normalization.
Spotify Premium listeners also have the option to change the normalization level. LOUD is -11 LUFS. Normal is -14LUFS. Quiet is -19 LUFS.
When LOUD setting is used and the original level of the master is under -11 iLUFS, it will be turned up regardless of it's maximum True Peak. They will apply a limiter to prevent distortion and clipping. There is no mention of what limiter they use and there is no way of previewing what effect it would have on the audio.
Less dynamic masters with a smaller iLUFS may be more liable to distortion when transcoded.
To check this, a mastered file can be auditioned in the codec that is used by the service before release. You will need a plugin for this. (Sonnox, iZotope, Apple)
For Spotify these are as follows.
  • Ogg/Vorbis (96, 160, 320 kbps)
  • AAC (128, 256 kbps)
  • HE-AACv2 (24kbps)

CORRECTION EDIT: I have removed the information about Spotify applying a limiter when turning up very quiet music as they no longer do this. Very quiet music 'is' turned up (if loudness normalization is active) but it will always have 1dB headroom. Limiting is only applied when in the LOUD setting (-11 iLUFS) for premium listeners. Sorry for any confusion.
 
Last edited:
I produce lots of very quiet music, and Spotify leaves my stuff pretty much intact. This might be a totally quiet track, or a track that has a modest level, then a quiet passage then medium again - none of my music has been mangled in any way.
 
I produce lots of very quiet music, and Spotify leaves my stuff pretty much intact. This might be a totally quiet track, or a track that has a modest level, then a quiet passage then medium again - none of my music has been mangled in any way.
That's great Rob!
None of the music I work on or have had uploaded to Spotify falls into the very quiet category, so I haven't been able to confirm how the normalization process would behave.
Hopefully, as in your case, it leaves everything intact and is a relatively non destructive process.
 
Last edited:
I produce lots of very quiet music, and Spotify leaves my stuff pretty much intact. This might be a totally quiet track, or a track that has a modest level, then a quiet passage then medium again - none of my music has been mangled in any way.
Hi Rob.

I actually got the information wrong about very quiet music as it was outdated. Spotify DOES NOT apply a limiter when turning it up now. They only use a limiter when the LOUD setting (-11 iLUFS) is used by premium listeners. I corrected my original post with the right info. Sorry about that.
 
Last edited:
Your post did make me thing - the same tracks are on youtube - put there by the aggregator, not me and comparing Youtube and Spotify - Youtube has clearly adjusted the levels, but I think it's mainly normalisation rather than compression. When I get time, I'll record them direct from the computer and see if they overlay in the editor - that would be interesting to check
 
I checked the stats and Youtube doesn't turn up quiet songs. It only turns down by normalizing to -14 iLUFS. You can check in the metadata directly on your Youtube video how much they have turned the audio down, or if it is under their normalization level.

Right click on the video as it plays. Select stats for nerds. In the Volume/Normalized category it will tell you the content loudness number. This number is the amount in dB your track has been turned down. If it is a negative value then it has not been adjusted and should be the original as uploaded.
 
Back
Top