[quote=“Coquet-Shack, post:21, topic:1424”]
Normalisation simply means turning all volumes up.
[/quote]We’re talking about loudness normalisation, not normalisation. If you don’t know the difference, you should read up on it, especially if you’re offering mastering services. My dissertation explains it all in detail.
You have called it a ‘normalisation boundary’ in your title, which is incorrect. Spotify has reduced its loudness normalisation target.
[quote=“Coquet-Shack, post:21, topic:1424”]
The increase in volume on the major platforms DOES NOT involve an crushing up of levels within headroom, it utilises integrated values which allows for peaks going OVER the stated LUFS lines.
[/quote]Completely incorrect. In fact that doesn’t even make sense.
When a recording is below the target LUFS, the PLR (peak to loudness ratio), the ‘new’ dynamic range, is reduced through compression/limiting in order to make the recording louder, thus reaching the target loudness.
Check Ian Shepherd’s page, there is a nice chart which explains it all:
Please correct me.
LUFS = an integrated loudness measurement?
If that is true, a piece can have peaks above and well below the LUFS target and still meet it?
So is the PLR relevant in that measurement?
I read Ian’s piece. But Spotify contradicts what he says, saying that it doesn’t apply limiters. I’ll try to find the doc they sent me.
PS: The increase in volume on the major platforms DOES NOT involve an crushing up of levels within headroom, it utilises integrated values which allows for peaks going OVER the stated LUFS lines. was REALLY badly worded. My apologies.
[quote=“Coquet-Shack, post:24, topic:1424”]
LUFS = an integrated loudness measurement?[/quote]Correct
[quote]If that is true, a piece can have peaks above and well below the LUFS target and still meet it?
[/quote]Correct. Integrated loudness is an average for the whole recording.
[quote=“Coquet-Shack, post:24, topic:1424”]
So is the PLR relevant in that measurement?
[/quote]Not entirely sure. I’m guessing it plays a part. The main pont is that PLR (similar to dynamic range) is reduced by limiting in order to make the track louder .
[quote=“Coquet-Shack, post:24, topic:1424”]
The increase in volume on the major platforms DOES NOT involve an crushing up of levels within headroom, it utilises integrated values which allows for peaks going OVER the stated LUFS lines.
[/quote]This is two separate things. The first part is talking about whether a limiter is used or not on tracks that are quieter than the target loudness, the second is simply a explanation of how the measurement is done.
Spotify can say what it likes. If you submit a -16 LUFS recording and it is played back at -14 LUFS then it is without doubt compressed/limited, (or perhaps clipped) - unless you or Spotify have knowledge of some hitherto unknown method of making tracks louder.
OK:
What happens if you simply raise the volume (turn up the gain on the 2 buss) on the -16 lufs piece? Could you get it to register -14lufs by doing that?
I’ve emailed the guy at Spotify to send that PDF again. I’m sure it said they would “normalise” by raising volumes but I’m starting to doubt myself because of your statements.
The masters I’ve done thus far have met the required LUFS levels with no problem. (Barring spotify which had a ludicrously high level - and I didn’t try for).
Assuming the recording is normalised to 0db then it’s going to clip by 2db. You can’t simply take up a fader to make a track louder, if you could we would never have bothered with limiters.
[quote=“Coquet-Shack, post:26, topic:1424”]
The masters I’ve done thus far have met the required LUFS levels with no problem. (Barring spotify which had a ludicrously high level - and I didn’t try for).
[/quote]That was the point of my first question. There aren’t any ‘required’ LUFS levels. It’s enough to be aware of target loudness levels and utilise them as a cue for your mastering loudness decisions, that is all you need. There’s nothing wrong with mastering louder than target levels, all that will happen is that the service will turn it down. As long as you have a decent dynamic range your recording will not suffer by simply being turned down. Conversely, it is important to remember that CDs (especially cars) generally do not have loudness normalisation, so you either need two separate masters for each track, (one loudness level for CD, one for streaming), or you choose a one-size-fits-all loudness.
To be honest, mixes seem to start falling apart when they are mastered any quieter than -12 LUFS in my experience. Loudness normalisation is (almost) all about turning recordings down, very few are too quiet and need boosting. Hours of experimenting has led me to a ‘go-to’ level of -10 LUFS, this seems to suit most genres. I have just done a CD at -12 LUFS but the artist was very concerned about maintaining dynamic range, and -12 LUFS was the best solution in this case.
In fact this led to a very interesting test, I mastered a track twice, the first at -10 LUFS and the second at -12 LUFS. Then I imported both tracks into a project so that I could A/B them, and dropped the -10 LUFS version by 2db on its fader (‘manual’ loudness normalisation). Result? A very subtle difference, noticeable the most on the ride cymbal.
Yep, “required” was my bad. What I mean is that I aim at what the platforms want. I’d rather get my masters to their targets my way than risk them effing about with them. Personal preference, I guess).
On mixes being turned down: my fear (and it’s shared by Shepherd) is that if you mix loudness in (I’m talking loudness wars crushed type loudness) and you bust the platform’s “Lufs level” (apologies my terminology sucks) what you’ll end up with is a crushed track that sounds softer than you planned. (I THINK that’s kinda what the pic shows) + the pic also says spotify uses limiters to adjust levels, I stand corrected).
Your experiment is one I must try. What was the difference? tonality? Perceived loudness?
I’m all for loudness normalization, but man, I was listening to spotify yesterday, and two songs came on in a row, both by the same band but from different albums, and the difference in loudness was not subtle. So what’s the deal? It’s not like these songs are dynamic by nature making it hard to level match. If any two songs should easily be able to be loudness matched, it’s these two. Is it due to the fact that LUFS aren’t that great a way to measure loudness? Is it a bug in Spotify?
Well, exactly. That is the whole point of loudness normalisation.[quote=“Coquet-Shack, post:28, topic:1424”]
Your experiment is one I must try. What was the difference? tonality? Perceived loudness?
[/quote]
Definitely not perceived loudness - again that’s the point of loudness normalisation.
The best way I can describe it, is that sonic bowl of porridge we all recognise as a hyper-compressed snare. There was a suggestion of that type of thing going on with the ride cymbal.
Could be a bug, could be anything, who knows? FYI Spotify doesn’t actually use ITU BS 1770 (LUFS algorithm). It uses…well, no-one actually knows what it uses.
