Mastering with LUFS meter...please advise

Hey All,
In the past I’ve just used the db meter and got it close to 0db because everything was CDs.
But now with streamers wanting specific LUFS numbers I’ve begun using this measurment.
Correct me if I’m wrong, but it seems that a best practice would be to create separate -16LUFS and -14LUFS files and, say, a -1db file for the CD.
Is this about right thinking?

  1. I think you are mixing up the concept and use of some of these terms. LUFS (in a nutshell) refers to average loudness. 0dB, -0.2dB, or -1dB typically refers to the peak levels. Apple and oranges. Even before the days of streaming, CDs had LUFS levels, we just didn’t know it yet :slight_smile:

  2. I disagree with making multiple loudness versions for various streaming services and CDs. The streaming services do the playback normalization so you don’t have to. Find an average loudness that works best of the material and move on.

If you do a more conservative master of typical modern pop/rock etc. at -14LUFS, and then a louder “CD era” master at say -9LUFS, the character of the two versions will be pretty different. The balance between melodic and transient info especially. It’s too confusing for the client/artist and the critical listeners.

Don’t overthink it, just know that once you reach a certain loudness point, most streaming services will start to turn it down and you get diminished returns.

This website may help you find the sweet spot for what works in a loudness normalized environment vs. a non-normalized environment such a CDs, and SoundCloud:

However, don’t rely solely on the numbers. Use your ears and let the numbers confirm what you are hearing and guide you if needed.

I agree with Justin’s observations.

I add that I am yet to receive any request from a label or artist management for anything other than a CD master, hi res MFiT and … in some cases … mp3s.

I say just make the best sounding master you are capable of, and let the streaming sites turn it up or down or do whatever they want. If you make something sound the best it can possibly sound, there is no reason to change it and make another version that sounds “different/worse”. Extra headroom and larger dynamic range is not always better. Sometimes a smaller dynamic range actually sounds better. Maybe just find the sweet spot where you’re happy.

Spend some time analyzing some of your favorite records and see where they are for LUFS – then you’ll know your personal preference for loudness and where you’d like to put things.

A good way to test this is to make a -16 and a -14. Then open iTunes and turn off Sound Check (iTunes’ normalizer). Put your -16 in a playlist and compare it to some tracks in the iTunes Store. The problem is if you submit at -16 it’s always going to be -16 whether normalization is on or off in the player, whereas most of the pop tracks in the store are submitted somewhat near the CD level most likely around -8 or -7 lufs, and they’re only normalized down to -16 if the user happens to have Sound Check turned on.

Same thing in Spotify. Add your -14 as a local file and turn off normalization in Spotify and compare. Your -14 will always be -14, versus most of the other tracks, which are probably louder and only normalized down to -14 when normalization is turned on in the Spotify program.

A good way to get familiar with mastering levels and online platforms is The Mastering Show podcast by Ian Shepherd. Especially this one, which is exactly about the topic of this thread:

Also, I’d be very wary of admitting something as loud as -8 LUFS to such a platform. Yes, they’ll simply turn it down - but the result is that all dynamics are unnecessarily lost, making your track sound flat and uninteresting against the rest.

Yes, I was wrong in mentioning a specific LUFS number at all, but that number is not unusual. My point was the files for online should definitely not be submitted at the platform’s normalization target of -14 LUFS or -16 LUFS simply because that’s their normalization target. Especially if it’s not even close to the level of the CD. In general the files should be close to the CD level (whatever that may be) and the adjustment to target only done by the end player. And that’s because normalization is probably off in half of the players out there, and on in the other half.

Thanks Guys. This was all very informative and helpful. Will follow this advice.

Good point. In my opinion in this context, the streaming loudness wars are still slightly on, since many listenders are going to listen/compare without Sound Check turned on or without the loudness normalisation in Spotify. It’s only You Tube and Pandora where loudness normalisation is on all the time.

Agree with the advice given on this thread and I think the loudness penalty site is definitely helpful. Although it’s not recommended to aim for a specific LUFS value all the time I think it depends on the context. When mastering a whole album it’s not healthy to aim for a specific fixed loudness target for each and every song (the loudness depends upon the musical or other context of the material) but you may want to have a rough loudness target for the album as a whole, and general guide loudness tolerances for each song. In combination of course with using ears.

I agree with Arjan that in general very loud hyper compressed masters may suffer if there is loudness normalisation applied by the streaming service. The loudness penalty site allows you to try it and see how much it’s likely to suffer.

Plus the fact that no one is delivering separate -16 or -14 masters for streaming when their CD is -10 or -11 LUFS. No commercial releases I know of from any of the mastering houses. If you do that you’re turning it down 5 or 6 db for no reason and your record will be that much quieter than everything else out there in players where normalization if off. Most importantly, for no reason at all. I’ve looked at streaming files vs cd from many places and it’s simply not done.

I don’t know where this idea started to do that, but I’ve seen it said elsewhere. But it doesn’t make sense.

And I think everyone’s advice not to squash things is good advice. To the op, If by db meter you mean VU meters, I think VU meters with needles often serve well as a 2nd reference.

Personally I see this from a different angle. I actually wish that all streaming services would engage the same industry standard loudness normalisation with no option to switch it off. Then adhering to a loudness standard within certain tolerances might make more sense. It seems to me that the current loose status of the streaming services makes a mockery of any kind of standardisation. But I may be missing the point.

I agree. They should pick a number and a method of measurement and be done with it. But I wouldn’t hold my breath. My point, if I had one, is that it’s crazy how many different ways a record can appear on different devices in different apps with different settings with different plugins (if you can call a normalizer a plugin). And the fact it’s a moving target with software updates. iTunes Sound Check normalizer didn’t used to affect previews in the store. Now it does. Big difference. Spotify local files aren’t affected by the normalizer at all, but streaming files are. Videos in Spotify are not affected by the normalizer. Spotify Web Player doesn’t have any normalization at all (last time I checked anyway). So my point was, the advice I’ve seen elsewhere to submit masters at -14 for Spotify, and -16 for iTunes is (in general) bad advice, because it doesn’t take into account the many cases where normalization is not engaged at all, often by user preference.

But there are always exceptions and it depends on the type of music and what sounds good, and what’s standard, and a lot of other things. But submitting -16 to iTunes just because it’s their normalization target is not realistic. The adjustment to -16 LUFS is really the job of the normalizer in the players.

Yes. Much of the advice from reputable sources suggest that it’s a good idea to submit material according to the respective loudness targets and suggest there is no point submitting masters louder than the loudness normalisation target for each service, with the assumption that loudness normalisation will be engaged all the time. It would seem it is only in You Tube and Pandora where loudness normalisation is always on. And the quoted target figures are not necessarily accurate / set in stone since each service uses proprietary loudness normalisation techniques which are not publicly shared and are subject to change. So the advice can be misleading and I’m sure on occasion many of us have been misled in one way or another. I noticed that Wavelab has MetaNormalizer (clip based) presets for ‘Spotify -14LUFS’ and ‘Apple Music -16LUFS’.

Well said, and good point about the presets in the metanormalizer. I guess I’ve thought of the Spotify and Apple presets there as useful for previewing, comparing, and testing, but not for delivery. I’ve only been asked to deliver to a specific target level twice, and both times it was for broadcast level -23 LUFS +/- 1 LU with max TP of -1 dbfs. But that was for a broadcast spec version of material related to a DVD Blu-ray release, not general music release. The metanormalizer was indispensible for that, but for iTunes and Spotify music delivery I would never normalize to the -14 and -16 levels just because it’s their nomalization target. As Rat said, CD, MFiT, and that’s basically it. Which of those they use for the video is anybody’s guess. Labels are not going to pay for yet two or three more masters, that would increase costs all down the supply line, and I don’t think such a thing has ever been done for commercial releases. Plus it doesn’t make sense when normalization can be and is turned on or off in players. The Spotify Web Player is the most surprising thing. No normalization at all, with no option for normalization, so for relatively loud popular music of many genres, a -14 LUFS master in Spotify Web Player will probably sound pretty quiet compared to most commercial stuff there.