PreSonus Blog

How to Make Spotify Happy

 

With physical audio media in its twilight, streaming has become the primary way to distribute music. A wonderful side effect has been the end of the loudness wars, because streaming services like Spotify turn levels up or down as needed to attain a specific, consistent perceived level—squashing a master won’t make it sound any louder.

However, the “garbage in, garbage out” law remains in effect, so you need to submit music that meets a streaming service’s specs. For example, Spotify prefers files with an LUFS of -14.0 (according to the EBU R128 standard), and a True Peak reading of -1.0 or lower. This avoids adding distortion when transcoding to lossy formats. If the LUFS reading is above -14.0, then Spotify wants a True Peak value under -2.0.

Fortunately, when you Detect Loudness for a track on the mastering page, you’ll see a readout of the LUFS and LRA (a measure of overall dynamic range), as well as the True Peak, RMS (average signal level), and DC offset for the left and right channels. Fig. 1 shows an example of the specs generated by detecting loudness.

Figure 1: Although the LUFS reading meets Spotify’s specs, True Peak doesn’t, and the RMS value of the left and right channels isn’t balanced.

 

 Note that this hits Spotify’s desired LUFS, but the left channel’s True Peak value is higher than what’s ideal. This readout also shows that the average RMS levels for each channel are somewhat different—the left channel is 1.2 dB louder than the right one, which also accounts for the higher True Peak value. This may be the way the artist wants the mix to sound, but it could also indicate a potential problem with the mix, where the overall sound isn’t properly centered.

A simple fix is to insert a Dual Pan into the Inserts section. Use the Input Balance control to “weight” the stereo image more to one side for a better balance. After doing so and readjusting the LUFS, we can now give Spotify exactly what it wants (Fig. 2). Also note that the left and right channels are perfectly balanced.

Figure 2: The True Peak and RMS values are now identical, so the two channels are more balanced than they were without the Dual Pan.

 A Crucial Consideration!

You don’t want to mix or master based on numbers, but on what you hear. If you set up Dual Pan to balance the channels, make sure that you enable/bypass the plug-in and compare the two options. You might find that balancing the left and right channels not only accommodates Spotify’s requirements, but improves the mix’s overall balance. If it doesn’t, then leave the balance alone, and lower the track’s overall output level so that True Peak is under -1.0 for both channels (or under -2.0 for LUFS values above ‑14.0). This will likely lower the LUFS reading, but don’t worry about it: Spotify will turn up the track anyway to reach -14.0 LUFS.

Coda: I always thought that squashing dynamic range to try and win the loudness wars made listening to music a less pleasant experience, and that’s one of the reasons CD sales kept declining. Does the end of the loudness wars correspond to the current music industry rebound from streaming? I don’t know… but it wouldn’t surprise me.

  • Interesting info there Craig – especially about Spotify adding the limiter if they raise the level. Makes sense of course, but the lesson there is – make sure you master loud ENOUGH. I remember being horrified the first time I read about EDM being mastered at -5. As a matter of interest I tried one of my tracks at -5 and it sounded horrendous! That’s an area where they really could benefit from taming the loudness war, although with no “normalisation” in the clubs (apart from riding the fader), I’m not sure where they’d start.
    Back to Spotify – I’m 100% in favour of normalisation but I do find it slightly unfair that they ask us to “keep it down” while they still have delivery platforms which don’t normalise.

  • Craig Anderton

    That may be due to Spotify asking artists to master to a uniform level, so most people are?

  • Frans van Nispen

    I never knew Spotify had a setting to enable or disable normalisation. But I just tried it and it doesn’t appear to do anything if I disable it.

  • Craig Anderton

    Thanks for the additional info! I’ve used Spotify on Android, iOS, and Windows app, so I didn’t realize the web player (and integration in 3rd party apps) had normalization turned off. However, the Spotify site says it’s “currently” turned off, so that might change. I also found out that when lowering a high LUFS, it’s just a volume change but when raising level, they add a limiter. I agree with you about rock – restricting dynamics can give a particular effect that persists even if the level is turned down. Then again, -9 is still reasonable. It’s the stuff that sits at -6 or even -5 that sounds “small” when the volume is normalized.

    FWIW with Pandora and YouTube, normalization is permanently on in Browsers, on by default with Amazon Music, Tidal, and Deezer (although you can turn it off in Tidal on iOS and Android), and off by default with Apple music. I’m not sure who’s doing album normalization vs. individual songs. Also LUFS readings aren’t always perfect.

    As to people saying music sounds “different” when normalized, it really is just a level change when going from higher to lower, so I suspect any tonal difference may be due to the louder level hitting a sweeter part of the Fletcher-Munson curve.

    Like you, I do master a little “hot” for some material because of how it affects the sound. On balance, though, I think the advantages of mastering to a consistent level – while imperfect – outweigh the disadvantages of having loudness wars, where people were destroying dynamic range because of needing to be “competitive” with playlists. If the reason for squashing is more volume, you can always get more volume with unsquashed material by turning up a volume control. But with squashed recordings, you can never get the dynamics back, regardless of level.

    Interesting discussion, and I’m sure more changes are in store.

  • Steve Mann

    Thanks for the feedback Craig. Regarding the true peak I usually leave 1dB headroom to account for transcoding. If I compare my -14LUFS masters with my -9LUFS masters with volume compensation so they both sound the same loudness, I like the added “pressure” I get from Ozone’s Maximizer when set up correctly. But this is of course just my taste and is probably more relevant in my case as I produce mostly rock music. But I do generally find that the Maximizer enables me to push the audio up to -9LUFS without any real detriment to the way transients (e.g. snare) are perceived and without any feeling that the dynamics are being overly squashed.

    Personally I think it’s dangerous to assume that Spotify listeners have normalisation switched on. I have read plenty of forums where people are sure that the normalisation changes the sound of the music and so switch it off. The Web Player also seems to have no option to switch normalisation on or off and it seems to be off by default.

  • Craig Anderton

    Spotify’s player defaults to “Normalize Volume,” but yes, you can defeat that under “Settings” to hear the music without volume normalization. I choose to normalize the volumes, and I assume most other people do too but I don’t have any stats of that. The advantage of catering to those who have chosen to normalize is that heavily-compressed albums can sound “smaller” when volume-normalized due to the lack of dynamic range. So I guess it comes down to whether you think more people listen to Spotify with volume normalization turned on or turned off, and which group is the target for your music. (FWIW I did a test on YouTube comparing my albums, which span four years and were mastered to different LUFS levels. They all have the same perceived level.)

  • Steve Mann

    Over the past year or so all my releases on Spotify have conformed to the requested -14 LUFS loudness, which is 5 dB quieter than I master for CD. I have recently discovered that the Spotify Web Player does not seem to equalise the perceived loudness between tracks – they all seem to be at their original levels. So all the releases of my peers who have simply uploaded CD masters sound way louder (sometimes nearly twice as loud) than my releases which sound very weak by comparison. I will never again upload to the streaming services at anything quieter than -9 LUFS.

  • Craig Anderton

    Thanks! Being a musician is where I got started 🙂 I’m putting the finishing touches on my 2020 album project, which was done entirely in Studio One. It will be posted on my YouTube channel soon.

  • I. C. All

    I’m glad to see the coming of the end of the loudness wars myself. It forces you to ruin the dynamic range and that means the song won’t have the intended effect on the listener. I’ve self-released a few CD’s (admittedly the first ones weren’t mixed/mastered ideally, due to limited experience) and I refused to squash them just for the sake of loudness. I even made up a little badge I put on the covers, “ALWR”, which stood for ‘Anti Loudness War Recording” LOL

  • Jeff Clark

    Thank you Craig Anderton! You have been chock full of good information over the years. It wasn’t until recently that I found out you are actually an artist too LOL. 😁

    Thanks again,
    Jeff Clark