[UPDATE: TIDAL now also use normalization, and Soundcloud say it’s “on the list”]
We know now that all the major music streaming services are using loudness normalization – higher level songs are turned down, aiming for a “target” loudness, which is different for every service.
Loud songs are turned down, quiet songs may be turned up on some services – IF there’s enough peak headroom.
And because the target level for some platforms is pretty loud, that’s a very significant “IF”.
YouTube and TIDAL don’t turn quieter songs up at all. And even on services that do, when there isn’t enough headroom to lift the level up without clipping, your music either won’t get turned up and will sound quieter than everything else as a result, or it may have extra peak limiting applied to get it up to the target level. Which may or may not sound good.
None of these is an ideal situation !
How loud is loud enough, and how loud is too loud ?
The answer is… it depends.
It depends on your views on dynamics, your views on the loudness war, and which platform we’re talking about – YouTube, Spotify, TIDAL – wherever.
To try and shed some light on the subject, I’ve put together an infographic summarising as much as I can about the topic – take a look, I hope it’s helpful !
(I’d like to say Thanks to members of my Home Mastering Masterclass course, who helped me get this infographic right, offering loads of helpful feedback and suggestions.)
Click on the image to see a high-res version, or click here for a PDF copy you can download and print.
If the terminology used is new to you (LUFS, PLR etc) please see the end of this post for a brief glossary.
Edit – for those who have been asking, BeatPort and Pandora also use a loudness target to even out replay volume. SoundCloud and Bandcamp currently don’t, however in my opinion it’s only a matter of time. And, I’ve been right about this before.
So how does it sound ?
In all cases except Spotify, it doesn’t have a sound. There’s no extra processing used, it’s just a simple replay level adjustment, which works reasonably well, in most cases.
The catch, as you can see from the infographic, is that if your music is too dynamic, it may not be turned up as much as other songs – like the blue song playing on YouTube in the infographic, for example.
Whereas on Spotify, this song would be played at the same level as everything else, but it would have extra limiting applied, which may not sound great.
Not happy ?
You might well be looking at this infographic and saying – “This is crazy ! Why shouldn’t I master my music at really high levels, if I want to ?” – or alternatively, “I love dynamics ! Why should my music be penalised for being less squashed than YouTube or Spotify want it to be ?”
And you’d have a point, in both cases. The answer is, that large changes in loudness are the number one cause of complaints from listeners, so loudness management of some kind is inevitable from this point on. The best we can do is try to encourage streaming services to implement it in the best way possible.
A while ago, Spotify removed the preference setting that allowed you to disable loudness management, to avoid over-limiting dynamic music. But after pressure from users, they brought it back.
We can do the same for loudness management. Spotify have shown they listen to their users before, and maybe they will again – you can add your name to the list asking them to reduce their playback level here. Apple’s Sound Check is pretty good already – they’re the only platform whose target loudness complies with the AES Streaming Loudness Guidelines – we just need them to pressure them enable it by default. And YouTube ? We’re working on it…
The use of loudness management by all the major streaming services is the first step in the right direction. Hopefully before long, replay loudness online will be standardised at a sane level like Apple’s Sound Check, and we can go back to mastering our music exactly the way we want to !
If you’d like to see this happen too, please sign the petition.
In the meantime…
Until then, the good news is that you can optimise your music today so that it plays consistently on all platforms – and it can still sound great.
In a nutshell:
You need to keep the peak to loudness ratio (PLR) of your music low enough to fit into the “loudness space” allowed by your chosen streaming platform, as shown in the graphic above.
Meanwhile, keep the minimum short-term peak to loudness (PSR) of your music above 8, to make sure your music isn’t turned down too much. My Dynameter plugin was designed to make both these steps as easy as possible – for more info, click here.
Some of the terms and topics in the infographic may be new to you – if so, this should help get you up to speed:
LUFS – Loudness Units (Full Scale)
The internationally agreed method of measuring loudness, measured in loudness units LU). Loudness units take into account the fact that our ears are more sensitive to some frequency ranges than others. How do LU relate to more familiar measurements, like RMS, dBFS etc ? See here:
dB TP – True Peak Level
Normal peak meters only read sample levels, but in some situations the decoded digital signal can generate levels which exceed 0 dBFS, especially if the music is mastered or mixed at a very high level.
“True Peak” meters use oversampling in order to measure these higher values, sometimes giving results as much as 3 or 4 dB above 0 dBFS ! Lossy encoding for streaming or mp3 can also generate extra peak information above zero, so I recommend you don’t exceed -1 dB TP when mastering.
More information in this video.
PLR – Peak to Loudness Ratio
The difference between the peak level and the Integrated loudness in LUFS. “Integrated” loudness is an overall figure for a whole song or piece of audio. PLR gives an indication of the overall “crest factor” of the music – how compressed or dynamic it is. For a great 2-minute visual summary of this, click here:
So for example, if the integrated loudness is -15 LUFS and the music peaks at -1 dBFS, the PLR is 14. If the music peaks at -1 dBFS but the integrated loudness is -10 LUFS, the PLR is only 9, which means it has probably been quite heavily compressed and limited.
PLR values measure something similar to the “DR” values you may be familiar with from using the TT Meter, but typically read a few points higher – for example PLR 11 is roughly equivalent to DR 8.
The short-term peak to loudness ratio is called the PSR, and is much closer to the TT Meter’s DR readings. It’s a great way to assess the dynamics of your music, and is displayed by my Dynameter plugin – for more info, click here.