How Loud is Loud Enough?

  • 31

Baphometrix explores the topic of loudness so that you can make better decisions when mixing and mastering.

Baphometrix is a Producer-DJ specializing in festival-oriented bass music genres (with some hip-hop on the side). She is a student of ill.Gates and a member of Producer Dojo’s Class of 808.

In this article, I shed some light on a subject that can be very confusing to newer producers in electronic dance music genres: loudness. Specifically, “how loud should I make my masters in this day of normalized streaming platforms and easy self-publishing?” I see variations of this question pop up time and again in the producer communities I’m part of. I also struggled with this same question for a long time.

I’m going to start with four assertions that might surprise you, but stick with me before concluding I’m wrong:

1 – All of the dynamic-range experts/advocates are wrong when it comes to electronic dance music

2 – A fair bit of the “professional mastering engineer advice” floating around is wrong when it comes to electronic dance music

3 – Some of the “all in one” mastering plugins out there won’t allow you to hit useful loudness targets for electronic dance music

4 – More than a few of the “how to master your songs for Spotify and streaming platforms” essays floating around are wrong when it comes to electronic dance music

The argument for dynamic range overlooks two important factors for electronic dance music

Let’s start by reviewing the general argument for using the full dynamic range of the normalized platforms:

Assertion 1 – More dynamic range = “punchier,” and more punchy is perceived as loud without being fatiguing

Assertion 2 – Less dynamic range = squashed, over-compressed, lifeless, dull, and fatiguing

These assertions are followed by the conclusion that you should always take advantage of the full dynamic range possible on the normalized streaming platforms, and therefore masters hitting roughly -14 LUFS integrated will always sound better and louder on those platforms.

This argument is usually backed up by charts like this one by Ian Shepherd. Ian is a professional mastering engineer, a well-known champion in the ongoing fight against the “loudness war,” and a strong advocate of “dynamic range is always better than squashed, hyper-loud masters.”

So here’s the thing: I agree fully with Ian’s viewpoint for a lot of musical genres, but not for electronic dance music. His argument for more dynamic range as summarized in that chart can seem very compelling, but it overlooks two important real-world needs and characteristics that are unique to electronic dance music.

1 – Electronic dance music is designed primarily to be mixed seamlessly in real-time with songs by other artists, and played through big sound systems. This means the spectral characteristics of all songs in a DJ set should be roughly similar, and the loudness to which a song is mastered imparts its own spectral fingerprint.

2 – Songs in most electronic genres are designed to move air and create a physical sensation for club- and festival-goers. These songs are not just passively listened to on Spotify (and other streaming platforms) through consumer-grade speakers, headphones, ear buds, and so on. To move air in this manner requires a heavy emphasis on sub frequencies, which in turn requires a stronger counterbalance in high-mid and highs for spectral balance. Broadly speaking, this means more overall density of sound, which by necessity creates reduced dynamic range. According to the dynamic-range advocates, the resulting music should sound squashed and lifeless, and yet this isn’t the case at all! While it can be difficult to get the right results, we do indeed hear loud, dense electronic music all the time that sounds and feels “punchy” and “vertical” AF.

Even on the normalized platforms, if you pay close attention you can hear that some songs still seem noticeably louder than others despite everything being normalized to roughly -14 LUFS. This happens because a well-made loud master is effectively very saturated in a way that takes advantage of Fletcher-Munson response, creating a heavier, fuller feeling of density at any loudness level, compared to more dynamic masters with less overall saturation.

The DJ dilemma

Any DJ or ProducerDJ who performs at clubs and festivals will also probably strongly disagree with the sentiment that “the most possible dynamic range is always better.” Instead, what’s important to a DJ–especially in the area of heavy festival-oriented bass music genres–is spectral consistency.

Here’s a thing I learned the hard way when I transitioned from simply being a Producer to being a ProducerDJ. When I first started performing and mixing live, and playing long sets with music from other artists mixed in with my own, I was using fairly dynamic masters for my own songs at first. (Based on the convincing arguments of the dynamic range advocates). I quickly learned that was a mistake. A serious mistake!

All the songs in my crates were normalized down to the same general loudness (roughly in the range of -11 to -12 LUFSi), yet for some reason my own songs still sounded quieter and more “wimpy” compared to the songs from other artists that I was mixing into and out of. Why? Because all the other songs were simply more saturated and dense because they’d been pushed harder into the final mastering limiter to hit competitive loudness war levels!  Their spectral balance was fundamentally different from my more dynamic masters!

As a result, I’d find myself reaching for the bass and midrange controls, and even after pushing those frequencies a little bit, still wishing I could just push up the gain a wee bit. But doing so would mean redlining into the front of house limiters, and that would have just killed the sound even worse.

This was my big “aha” moment. This is the aspect of loudness and the loudness war that is never really talked about in generic mastering advice or in the arguments of the dynamic range advocates.

Is the EDM loudness war “bad”?

Orchestras, jazz ensembles, world musicians, guitar/bass/drum-oriented bands in country, rock, alternative, etc… These are all organic instruments being recorded through analog gear. There are inherent properties and limitations in the analog signal path on its way to digital. And the original sound sources (the organic instruments) tend to have a huge dynamic range, which is therefore is an inherent part of their “sound.”

But electronic music? Or pop or hip hop? These genres all use heavily processed and saturated sounds throughout the entire mix. Probably the only organic instrument in any of these genres is the singer’s voice. And in these genres, even the singers are heavily processed and compressed to sit on top of an already dense and saturated mix.

The bottom line is that many electronic genres, and hip hop, and pop would not sound “right” if they were mixed and mastered to a level that makes full use of the dynamic range on normalized platforms like Spotify, Apple Music, YouTube, and so on! These genres are all meant to sound saturated and dense. And this is especially true of the bass music genres, where the sub and kick are extremely prominent and the rest of spectrum has to be balanced against that heavy bottom end.

Now, within the electronic genres there are varying degrees of loudness/saturation/density for different subgenres. For example, brostep and metalstep and doomstep are meant to be very aggressive and in your face. This is why you see ridiculous loudness numbers on some masters in these subgenres. If the sounds in the original mix are saturated to within an inch of their lives for maximum aggression, then the master itself just turns out loud AF. (And the mastering engineer might squash it extra hard to really bring out the edge and harshness.)

But even the more mellow/chill “electronica” genres are still fairly dense and saturated compared to most genres based on organic instruments! It’s just the nature of most electronic, hip hop and pop music to sound louder and denser and more saturated than other genres.

So at the end of the day, we could argue in the EDM world (and hip hop and pop world) that some masters are still stupidly louder than they need to be. We could argue that some loudness target like -8 LUFS integrated (LUFSi) across the loudest climax drop is “loud enough,” but it’s never going to happen. In the EDM world especially, producers (and engineers) delight in breaking the rules, and in the EDM world, even massive clipping distortion can sound “good” in the right context. Even clipping the master bus for the hardest distortion possible can be desirable for some songs!

Therefore, at the end of the day only one  “golden rule” really exists for loudness in electronic, hip hop, and pop music:

The Golden Rule – Your general loudness must be competitive with other songs a DJ is likely to play in the same set as your songs.

What exactly is “competitive loudness”?

Competitive loudness means that the overall spectral characteristics of your songs and the other songs in the same genre playlist or DJ set must sound essentially similar when all of the songs are normalized to the same general loudness. Notice that I’m stressing the spectral characteristics and not just a simple loudness value! In other words, what’s really important is that your overall saturation and density must be similar to other songs in DJ set or in that EDM playlist on Spotify or SoundCloud, etc.

The actual loudness itself (as measured in LUFS or RMS or DR) doesn’t matter that much, because:

1 – DJs either normalize their crates beforehand, or they manually ride volume during sets and therefore “manually” normalize every song.

2 – The major platforms like Spotify, Apple Music, Pandora, and so on normalize everything down to roughly -14 LUFS integrated (LUFSi) across the entire span of the song end-to-end.

3 – On the two remaining non-normalized platforms like SoundCloud and Beatport, listeners will reach for the volume knob and “manually” normalize every song that rolls across a playlist.

So this is why in electronic, hip hop, and pop genres, more than anything else you should be meticulously using reference tracks to fine tune the sound of your masters. To a large extent, you have to rely on your ears more than your eyes. Forget what the loudness meters are reading (to some extent). Instead, focus on the spectrum analyzers and spectrograms, and solo-filter chunks of the spectrum, and use your ears to ensure that the overall saturation and density and brightness and stereo field sound essentially similar between your track and your reference tracks.

At the end of the day, though, you’ll probably find that when you hit an “essentially similar saturation/density/brightness,” your loudness meters will be pretty close to the loudness meters of your reference tracks. Which brings us to a useful shortcut for achieving that “essentially similar” sound…

Spectral shapes and loudness targets are your best friend

Fortunately, it turns out that IF you mix and master to achieve a specific spectral shape and a specific loudness target, your music will naturally end up in the same ballpark as everyone else in terms of overall saturation and density! It’s actually very simple and boils down to three general rules of thumb:

Rule 1 – Your master should have roughly the same spectral shape as pink noise. In the bass music genres (or even hip hop and pop songs with a really heavy bottom end), everything from about 700 Hz and lower will sit up slightly louder than the pink noise slope (especially the sub range itself), but from about 700-1000 Hz and up your spectral slope should look really close to spectral slope of pink noise. And depending on taste and other factors, it’s normal and okay for the 16 Khz and higher range to start rolling off rapidly compared to pink noise. I can’t explain all the nuances of this here in this document, but these few guidelines should serve as solid clues for your own exploration. One simple tip is to use a spectrum analyzer that you can set to a slope/tilt/roll of 3.0 (or -3.0, depending on the specific analyzer). This makes it so that pink noise looks “flat across”–a horizontal line–in the analyzer. Make sure your master looks more or less “flat across” (with < 700 Hz sitting above the flat line and maybe > 16 Khz starting to slope below the flat line), and you’re in good shape. Tip: If you look at the screenshot at the top of this article, you can see the type of spectral shape I’m trying to describe.

Rule 2 – No more than 3 to 4 dB of gain reduction on your final mastering limiter (ideally only 2-3 dB of GR). Any more than this and even the best mastering limiters will start pulling the mix apart in undesirable ways. If you can set up your mix chain and earlier parts of your mastering chain to hit your desired loudness target with only 2 to 3 dB of gain reduction on the final mastering limiter, that’s about as perfect as you can ask for, and will add just the right amount of final saturation and density to everything that’s happened before this point!

Rule 3 – Make sure your master is hitting a loudness target somewhere in the range of -8.0 to -6.0 LUFS integrated (LUFSi) as measured across the loudest “climax” drop of your song. I personally shoot for -7.0 LUFSi in this measurement on nearly every song, and only settle for a lower value if my mastering limiter is doing more than 4 dB of gain reduction at the -7.0 loudness target. I almost never shoot for louder than -7.0, but I’ll accept anything down to -8.0 depending on overall style of the song itself.

You really don’t need to push for anything louder than -7.0 LUFS integrated across your loudest drop. Doing so will simply result in more harshness than necessary for electronic genres, and make it harder to acheive a “punchy” and “vertical” feel. -8.0 to -7.0 LUFSi is fully “competitive.” Your song will sound essentially the same as even ridiculous -3 or -4 LUFSi masters coming out of a DJ set.

What happens if you follow the three golden rules

If your loudest drop is hitting -7.0 LUFSi, then for most typical song structures, the entire song end-to-end will measure roughly between -7.2 to -9.5 LUFSi depending on how long and quiet your intros, breakdowns, and outros are.

It’s worth noting that an average reading of -7.0 LUFSi across your loudest drop will also hit maximum short-term LUFS (LUFSs) values of anywhere between -5.5 and -4.5 in many styles of heavy bass music. So that’s equivalent to occasionally hitting as high as -4 to -5 RMS in older-style RMS meters. It’s worth keeping an eye on your maximum short term readouts. If you see a max LUFSs value of -4.0 or higher, that’s an indicator that your track leveling is off and there’s one sneaky sound that’s probably still louder than it needs to be.

Most importantly, following these three golden rules will ensure that the overall saturation and density of your songs will be comparable to everything else on Beatport, which means you’ll mix well with other songs in a DJ set.

So what about different masters (and loudness targets) for different platforms?

Before I learned my hard lessons as a DJ, I made one loud master for SoundCloud and a different, more dynamic master for my distributor to all the major platforms (including Beatport). But no longer! I regret giving advice to this extent in the past!

Nowadays, I’m firmly in the “go loud or go home” camp, and the “one master to rule them all” camp. If you’re an electronic, hip hop, or pop producer, my firm advice is to master to a competitive loudness target as described above, and distribute that one master everywhere. It will simply sound better–everywhere, on every platform. And DJs will thank you.