Potential problem mastering for Spotify, Youtube etc.

[jmm55]

The song I'm working on is almost finished. I'm mastering it but will also have it mastered by a proper engineer. This heavy pounding rock song has three very distinct but complimentary parts, with a different beat, dynamics, sonic signature, and mix. The last part, arguably the focus of the song, is harder and faster than the first two, and there is a short pause between sections.

It has to be mastered from end to beginning so that the three parts balance out. This much I know even though I'm just a journeyman. I have the last part at -14 lufs and peaks of about -.5 dB, and I like it, but here's the problem. If I master the song such that parts one and two match part three in lufs, it's a disaster because the first two parts sound much louder in spite of what the meters say. The sweet spot for marrying these parts is with the first two down about 3 lufs.

If my mastering comes in at -14 lufs for the last part, I'm concerned that the leveling algorithms of Spotify, youtube, etc. might raise the level of the first two parts, which would certainly ruin the song. Is there any chance that their leveling algorithms might raise the level of the first two parts? Any thoughts on this?

[RichardT]

There's no reason why the three parts need to have the same LUFS. That's not necessary for them to hang together. Different arrangements usually sound best with different loudnesses.

You may need to adjust the LUFS of each section relative to what's best for each section individually to get the three pieces to cohere - that's normal. But let the music tell you what to do, so to speak.

At standard loudness settings (-14 LUFS) Spotify used to raise quieter pieces to -14 LUFS, by limiting them if ncessary. Now it will only do that if the quieter pieces peak below -1 dBFS, until the point where the peak reaches -1 dBFS, i.e. no limiting is applied.

This assumes you are releasing the music as 3 separate tracks. If it's going to be 1 track, then the relative dynamics of the sections will always be preserved.

At the -11 setting, Spotify does limit to achieve that level, and your relative dynamics will be changed. But there's no point worrying about that, it's completely outside your control and it happens to everybody.

[Wonks]

If it’s one song/one file, then any adjustment will be made to the whole file, not parts of it.

[The Elf]

I've often automated mastering settings to accommodate different parts of a song, so there's no reason your engineer couldn't do that.

[James Perrett]

Always let your ear be the final judge. -14LUFS isn't set in stone although the whole track will be set to around that level if it is played as part of a playlist. I would try putting all 3 sections together at the levels your ear says is right and checking the LUFS level of all three combined. You can then try bringing the level up via a limiter and see how loud you can make the end section without it sounding too mangled. I'll often find that rock tracks can go up to -12LUFS with virtually no limiting at all (just the odd peak brought down) so I would expect that you can bring your song up to -14LUFS overall with no problem.

[Stuart79]

What Wonks said.

-14 refers to integrated loudness, i.e. the loudness measurement of your track from start to finish. Your track doesn't have to be a static loudness throughout. Parts of your track can be louder than this, parts can be quieter. You can look at the short-term loudness reading to get a sense of how loud the track is as it plays.

As a guide, Ian Shepherd (Mastering Engineer) suggests getting the loudest section of your track to -11 (looking at short-term loudness) and then balancing the rest of your track to this musically, i.e. what sounds good to your ears in relation to the loud part. As your earlier parts will be quieter, the integrated loudness will drop down closer to -14.

[jmm55]

Thanks for all the helpful replies.

[Vlaaing Peerd]

Perhaps I don't fully understand the question, but to me it appears the TS is worried about how the media upload sites handle the 3 separate parts and adjusting it to the same average loudness, which breaks up the balance between these three.

If that is the case, by my knowledge you can only avoid it by mastering all 3 parts as one audio file instead of 3 separate ones. Otherwise, they would indeed be leveled to same levels of LUFS by media sites.