The post title says it all – in the last few days it’s become clear that Spotify have chosen to reduce their playback loudness reference level from approximately -11 LUFS down to approximately -14, broadly in line with YouTube and TIDAL.
This is a big deal, and in a minute I’ll discuss why, but before that – what does it mean, in simple terms ?
[This change is very recent, and you may need to update to the latest release of Spotify before you see it – the build number we are testing is 22.214.171.1249.g3809528e. It’s also possible this change hasn’t rolled out in all territories at the time of writing – 22nd May 2017]
In a nutshell, it means it doesn’t matter how high you push the level of your mixes and masters. Once the raw loudness of the files gets past a certain point, online streaming services will turn them down – keeping them all at the same reference level, to stop users being annoyed by sudden changes in volume.
Exactly where the “point of no return” is varies slightly between different streaming services, but Spotify always used to be the loudest, by a whopping 2-3dB.
And this was a real shame, because it put pressure on musicians, labels and engineers to make the raw loudness levels higher to try and “compete” – even if it didn’t suit the style of the music.
But now, all that has changed.
Why this matters
YouTube, Spotify and TIDAL all now use playback reference levels within a dB of each other, and Apple Sound Check and Pandora are another 2 dB lower than that, matching the recommendations of the Audio Engineering Society for streaming loudness.
So there’s no pressure any more to master louder in order to “compete” on Spotify – you can use the same guidelines for all the major streaming services, and be confident of a great-sounding result.
You can have great dynamics and sound loud – that’s a win-win !
How Loud ?
In a nutshell, the new magic number for Spotify is a reading of -14 LUFS integrated, meaning an overall value measured across the whole song, while keeping peak levels no higher than -1. Bear in mind that you shouldn’t regard this as a target. Spotify will adjust your music’s playback loudness to this kind of level, so it’s better to regard this is an opportunity to choose the loudness that you think sounds best for the material, without having to worry about “competing”.
YouTube’s reference level is actually 1 dB louder since the Spotify change, so you might choose to push things a little harder if maximum loudness on YouTube is important to you. If your music has varied dynamics though, it probably isn’t necessary.
And of course you do still need to keep an eye on the “crest factor” – the difference between the peak level and the short-term loudness. If this drops too low, your music may be turned down more than you expect. This value is labelled PSR in my Dynameter plugin, which was designed specifically to optimise this value property for optimal audio dynamics.
A huge improvement
This change is fantastic news. The -14 LUFS figure may not comply with the AES recommendations, but the reality is that this figure allows plenty of peak-to-loudness headroom for most mainstream music these days to have plenty of dynamics and sound great – which is a win-win for everyone.
I’ve been campaiging a change like this for some time now, both on the Spotify forums and via the Streaming Loudness Petition. There’s no way to know whether either of these initiatives actually influenced Spotify’s decision, but it really doesn’t matter.
The great news is that all the online streaming services now cater for music with decent dynamics – and they’re close enough to each other that there’s no need to create specially optimised masters for each platform – although this is still an option for people who want it.
Over 6 years ago now, I predicted that Spotify would end the loudness wars. Today is another important step towards that prediction coming true.
Hats off to Spotify, and long may the trend continue !
You may be reading this thinking – “What’s the big deal ? it’s just normalisation”.
And you’re right – this kind of processing won’t fix the damage that’s already been done in the process of making those loud-songs-that-are-being-turned-down loud in the first place.
But over the longer term, it removes the incentive to do it again. Sooner or later, the questions change:
Old question: “Why does Song X stand out on the CD changer ?”
Old answer: “Because it’s louder”
New question: “Why does Song Y stand out online ?”
New answer: “Because it has great dynamics”