Targeting Mastering Loudness for Streaming (LUFS, Spotify, YouTube)-Why NOT to do it

Music Computer
#1

Below I am sharing something that I send to my mastering clients when they inquire about targeting LUFS levels for streaming services. Months ago I posted an early draft of this in another thread so apologies for the repetition. I hope it is helpful to some readers to have this summary in it’s own thread. Discussion is welcome.

Regarding mastering to streaming LUFS loudness normalization targets - I do not recommend trying to do that. I know it’s discussed all over the web, but in reality very few people actually do it. To test this, try turning loudness matching off in Spotify settings, then check out the tracks listed under “New Releases” and see if you can find material that’s not mastered to modern loudness for it’s genre. You will probably find little to none. Here’s why people aren’t doing it:

1 - In the real world, loudness normalization is not always engaged. For example, Spotify Web Player and Spotify apps integrated into third-party devices (such as speakers and TVs) don’t currently use loudness normalization. And some listeners may have it switched off in their apps. If it’s off then your track will sound much softer than most other tracks.

2- Even with loudness normalization turned on, many people have reported that their softer masters sound quieter than loud masters when streamed.

3 - Each streaming service has a different loudness target and there’s no guarantee that they won’t change their loudness targets in the future. For example, Spotify lowered their loudness target by 3dB in 2017. Also, now in Spotify Premium app settings you find 3 different loudness settings; “Quiet, Normal, and Loud”. It’s a moving target. How do the various loudness options differ? - The Spotify Community

4 - Most of the streaming services don’t even use LUFS to measure loudness in their algorithms. Many use “ReplayGain” or their own unique formula. Tidal is the only one that uses LUFS, so using a LUFS meter to try to match the loudness targets of most of the services is guesswork. Edit: In September 2019 YouTube switched to -14 LUFS. In December 2020 Spotify began transitioning from “ReplayGain+3dB” to -14 LUFS. It’s good that a normalization standard is emerging. It’s also more evidence that they can change these targets at any time.

5 - If you happen to undershoot their loudness target, some of the streaming sites (Spotify, for one) will apply their own limiter to your track in order to raise the level without causing clipping. You might prefer to have your mastering engineer handle the limiting.

6 - Digital aggregators (CD Baby, TuneCore, etc.) generally do not allow more than one version of each song per submission, so if you want a loud master for your CD/downloads but a softer master for streaming then you have to make a separate submission altogether. If you did do that it would become confusing to keep track of the different versions (would they each need different ISRC codes?).

It has become fashionable to post online about targeting -14LUFS or so, but in my opinion, if you care about sounding approximately as loud as other artists, and until loudness normalization improves and becomes universally implemented, that is mostly well-meaning internet chatter, not good practical advice. My advice is to make one digital master that sounds good, is not overly crushed for loudness, and use it for everything. Let the various streaming sites normalize it as they wish. It will still sound just as good.

If you would like to read more, Ian Shepherd, who helped develop the “Loudness Penalty” website, has similar advice here: Mastering for Spotify ? NO ! (or: Streaming playback levels are NOT targets) - Production Advice

#2

Hey Justin,

Great post. For awhile I got on the bandwagon of multiple masters, one for CD, one for streaming. It was clunky, confusing to clients, aggregators wouldn’t receive both, more questions arose, etc. I have since moved back to a single file that is optimized mainly for CD while being auditioned as an mp3 to make sure there aren’t any nasty side effects of file compression to the track. Haven’t done any work that has requested MFiT, so I’ll cross that bridge when I get to it.

I think one of the fundamental issues here is the question of what is the purpose of making a track “loud”. I remember running into this way back with my mentor in live production. I was constantly questioned on my use of relatively heavy compression on numerous channels for rock and metal music (vocals, kick, snare, bass, etc). The challenge was that there was no need to reduce dynamics because of the excessive amount of available headroom. I couldn’t articulate my reasoning 14 years ago and I may not be able to today, but I will try.

I believe the goal here should be to find the appropriate “density” for the track. Rock or Pop tracks with full instrumentation mastered at an Integrated Loudness of -14 LUFS lack density and everything that comes with that (harmonics, distortion, saturation, pumping, balance shifts). Yes you can turn it up to sound as loud as a track with an Integrated Loudness of -7 LUFS, but it feels drastically different. Maybe everyone understands this already and I’m late to this experience this personal revelation.

Now I realize I was compressing tracks heavily in live production to create density and for shaping, not necessarily to prevent running out of headroom or for volume. While mastering, I compress, saturate, clip, and limit tracks fairly heavily to create density and for shaping and adding some amount of saturation and harmonic distortion. I find this more pleasant than the alternative of minimal to no clipping, saturation, limiting, etc. Who am I to go mucking up peoples well crafted mixes? I don’t know.

There are obvious examples of pushing tracks too far and the dangers of that. I am looking to strike the appropriate balance of perceived density for the track and not necessarily “volume” or “loudness”. Basically I’m trying to say, I agree.

Christopher

1 Like
#3

Hey Christopher,

Thanks!

Yes, density has become a desired sound quality for sure. In an ideal world only the sound (density included) of a track would matter, and once that sound was achieved we’d be free to then set the loudness as high as the peaks allow, or at -14LUFS or wherever. There’s nothing stopping us from simply lowering the gain of our dense tracks and leaving extra headroom.

My post is in part premised on my acceptance of the fact that the vast majority of clients are very concerned with not sounding too soft relative to their peers, so unfortunately loudness for loudness sake often becomes a factor beyond just sonics. In other words, sometimes the appropriate density alone isn’t loud enough for the client. It then becomes my job to achieve the client’s desired loudness while maintaining the sound (density included).

In practice that has led me to find ways to get pure loudness as transparently as possible. A side effect of that is that I can bypass my loudness processors and still maintain fairly similar density, so my non-brick-walled vinyl masters don’t sound too different from my brick-walled digital masters most of the time. Not relying on loudness processors for density frees me to set loudness apart from density concerns much of the time. At least that’s how I try to do it…

It’s not easy to word these thoughts. I hope I’m making sense…

#4

You suggested one master which I love that idea

or

Would you recommend a 16bit/44.1 master for cd?

and a 24/44.1 or 48 or 96 bit for Streaming, I-tunes etc?

#5

Thanks for the post. Absolute truth.

The one time I mastered to -14LUFS, the
client complained that it wasn’t loud enough.

Check any modern commercial CD release.
Every track comes in averaged around -11,
I’ve seen some averaging at -8!

I think we’re done with this stupidity regarding how loud to master things. Make it as loud as
you want as long as it sounds good.