Appendix B: Decibels and Loudness Reference

The decibel is the most useful number in this book and the most misunderstood. Chapter 1 introduced it as the way your ears measure loudness; Chapter 21 turned it into a gain-staging workflow; Chapter 33 used it to declare the loudness war over. This appendix puts the whole story on one page: what a dB actually is, the three "double" rules and what each is for, and the three different zeros — dB SPL in the air, dBFS in the file, LUFS in the listener's perception — that share the same suffix and confuse everyone for a year.

One thing to fix before any of the numbers make sense, because it's the single most common beginner mistake.

A Decibel Is a Ratio, Not a Number

A foot is a foot. A kilogram is a kilogram. You can hold one up. A decibel is not like that. A decibel measures a relationship between two values — it's always a comparison, never an absolute amount. "Turn it up 3 dB" makes sense. "This signal is 3 dB" does not, until you say 3 dB compared to what.

That's why every dB you'll ever meet has a hidden suffix telling you what the comparison is against:

• dB SPL — compared to the quietest sound a healthy young ear can detect (the threshold of hearing). This is loudness in the air.
• dBFS — compared to the loudest level your digital system can represent before it clips (Full Scale). This is level in the file.
• dBu / dBV — compared to a fixed analog voltage. This is level in the cable, the world of your interface and outboard gear.
• LUFS — compared to a standardized loudness reference, with your ear's frequency bias baked in. This is perceived loudness, what Chapter 33 lives in.

Strip the suffix and the number means nothing. When someone says "the vocal's at minus six," ask minus six what. Nine times out of ten the confusion in a beginner's session — clipping they can't explain, a master that's quiet on Spotify, a meter that's red when nothing sounds loud — traces back to mixing up two of these scales.

The decibel is a ratio for a reason: your ears are logarithmic. A sound with ten times the physical energy doesn't seem ten times louder — it seems roughly twice as loud. A sound with a hundred times the energy seems about four times as loud. The dB scale is built to match that compression, which is why the whole audible range, from a pin drop to a jet engine, fits inside about 130 dB instead of spanning a number with twelve zeros in it. The math is doing your perception's job for you.

🔄 Check Your Understanding

1. A friend says "my kick is at -3 dB." What's the one question you need to ask before that sentence means anything?
2. Why does the entire audible range fit inside roughly 130 dB instead of needing a number in the trillions?

Verify

1. Minus three dB relative to what? dBFS (file level), dB SPL (loudness in the room), or dB of gain reduction on a compressor are three completely different statements. The suffix is the meaning.
2. Because the dB scale is logarithmic — it compresses each ten-fold jump in physical energy into a fixed 10 dB step — and that matches how the ear compresses loudness. A linear scale would need about twelve digits to cover the same span.

The Three "Double" Rules

Three different dB numbers all get described as "doubling," and producers mix them up constantly because nobody tells them the three doublings answer three different questions. Here they are, with the job each one is actually for.

+3 dB doubles power. This is the one that bites you in the mix. Pan a mono track to center and the DAW drops it about 3 dB on each side, because dead-center the signal plays equally from both speakers and the power adds back up — that's pan law from Chapter 21. Sum two uncorrelated tracks (two different sources, two different takes) and the level rises about 3 dB, not 6, because their peaks don't line up. Two identical copies of the same file, though, add a full 6 dB, because every sample lines up perfectly. The gap between "+3 for different things, +6 for the same thing" explains half the surprises in gain staging.

+6 dB doubles amplitude. Push a fader so the waveform is literally twice as tall and you've added 6 dB. Each bit of bit depth from Chapter 2 is worth roughly 6 dB of dynamic range — that's why 16-bit gives you about 96 dB and 24-bit gives you about 144 dB. Bits buy headroom in 6 dB chunks.

+10 dB sounds twice as loud. This is the perceptual one, and it's the one that humbles you. To make a part subjectively twice as loud as the rest of the mix, you generally need it about 10 dB hotter — which on a meter looks enormous. It's why the loudness war was a sucker's game (Chapter 33): squeezing a master 3 dB louder buys a difference your ear barely registers, while costing you all your dynamics. Like all these numbers, "+10 ≈ twice" is a rule of thumb, not a law of physics; it shifts with frequency and listening level. But it's the right ballpark, and it's the reason "just make it louder" is rarely the answer to anything.

dBFS — The File's Ceiling Is Zero

Inside your DAW, the scale that matters is dBFS — decibels relative to Full Scale. Here's the rule that surprises everyone: 0 dBFS is the absolute maximum, the digital ceiling, and every legal audio sample sits below it. All your levels are negative numbers. There is no "+2 dBFS" on the way to the converter. Try to go past 0 and the samples have nowhere left to count to — they slam flat against the ceiling. That's clipping (Chapter 2), and it sounds like crackle, fizz, and a hard ugly edge that no plugin removes after the fact.

So the dBFS scale runs downward from a hard ceiling:

   0 dBFS ─────────────────────  THE CEILING. Past here = clipping. Hard wall.
  -1 dBFS ░░  mastering true-peak target (Ch 32/33) — leave the converter room
  -3 dBFS ░░░  loud master peaks
  -6 dBFS ▓▓▓▓▓  ← mix-bus headroom target before mastering (Ch 21)
                 ▒
 -12 dBFS ▒▒▒▒▒▒▒  hot individual track
                 ▒
 -18 dBFS ████████████  ← target AVERAGE for tracking & mixing (Ch 21)
                 │       "analog 0 VU" lives about here
                 │
 -24 dBFS ░░░░░░░░░░░░░░░  quiet, conservative average
                 │
 -60 dBFS ·  noise-floor neighborhood / fades / room tone
                 │
   -∞ dBFS  digital silence (no signal at all)

The dBFS headroom column: 0 is the ceiling, audio falls below it, and the room above your average level is the headroom you're protecting.

Read that diagram as a workflow, not a set of laws. You aim individual track averages around -18 dBFS and let peaks land somewhere around -12 to -10 — not because -18 is magic, but for two practical reasons. First, headroom: the space between your loudest peak and the 0 dBFS ceiling is the room mastering needs to do its job (Chapter 31). Hand a mastering chain a mix that already peaks at -0.1 dBFS and you've left it nowhere to work. Aiming the 2-bus at roughly -6 dB peaks before mastering keeps that room reserved. Second, plugins: many analog-modeled EQs, compressors, and saturators (Chapters 22, 23, 26) are voiced to "see" signal around the level where an analog console expected 0 VU — and that calibration sits near -18 dBFS. Feed them a hotter signal and they distort earlier than the designer intended; feed them a colder one and they barely engage.

Two honesty notes, both straight from the chapters. Modern DAWs mix internally in 32-bit float (Chapter 21), so a channel that momentarily runs "over 0" inside the box won't actually clip as long as you pull it back before it hits the converter or the print file — but that's a safety net, not a license to ignore your levels. And -18 dBFS is a habit, not a law. It's a starting point that keeps you out of trouble and keeps your plugins happy. Your ears, and the matched-loudness discipline from Chapter 22, make the actual calls.

🔍 Why Does This Work?

Why does the digital world put zero at the top while the SPL world puts zero at the bottom? Because each scale measures relative to its own most useful reference. In a digital file, the one fixed, unbreakable limit is the clipping point — the largest number the format can store — so engineers set that as 0 and measure how far below it you are. In the air, the one fixed reference is the quietest thing you can hear, so that becomes 0 and everything audible is a positive amount above it. Same word, opposite anchors, because "the most useful thing to measure against" is different in a file than in a room. Get clear on which reference you're using and the "all my numbers are negative / all my numbers are positive" whiplash disappears.

dB SPL — Loudness in the Real World (and Your Ears Are the Gear You Can't Replace)

Out in the air, loudness is measured in dB SPL (Sound Pressure Level), where 0 dB SPL is roughly the threshold of human hearing. This is the only one of these scales your body experiences directly, and it's the one with stakes that don't reset when you close the session: your hearing does not regenerate. You can upgrade your monitors, your interface, your room, your plugins. You cannot upgrade your ears. They are the one piece of gear in your studio with no replacement part and no warranty, and the entire premise of this book — trained ears beat expensive gear — collapses the day you damage them.

So here's the reference ladder. Treat the numbers as the widely-cited ballpark figures they are, not lab readings — actual levels vary with distance and source:

dB SPL │ Real-world reference                              │ Exposure reality
───────┼──────────────────────────────────────────────────┼──────────────────────
   0   │ threshold of hearing (silence you can just detect)│ ·
  20   │ ░ rustling leaves, a quiet recording booth        │ safe indefinitely
  40   │ ░░ quiet library, a hushed room                   │ safe
  60   │ ▒▒▒ normal conversation at arm's length           │ safe
  70   │ ▒▒▒▒ busy restaurant, a calibrated mix level       │ safe
  85   │ ▓▓▓▓▓ heavy city traffic from the curb            │ ⚠ ~8 hours, then risk
 100   │ ████ nightclub, loud live show, power tools       │ ⚠ minutes, not hours
 110   │ █████ front row at a concert                       │ ⚠⚠ damage builds fast
 120   │ ██████ threshold of pain, jet engine nearby        │ ⛔ immediate risk
 130   │ ███████ past pain — injury territory               │ ⛔ leave now
───────┴──────────────────────────────────────────────────┴──────────────────────

The dB SPL ladder: everyday sounds mapped to loudness, with the exposure zone where listening turns into damage.

The number to tattoo on the inside of your eyelids is around 85 dB SPL for about 8 hours — the commonly-cited threshold where prolonged exposure starts putting your hearing at risk. And it's not a linear budget. The rule of thumb is brutal: for roughly every 3 dB louder, the safe time roughly halves. So if 85 dB buys you about eight hours, somewhere near 88 dB it's about four, near 91 dB about two, and a 100 dB-plus club or show is doing measurable work on your ears in minutes. The damage is usually painless and permanent — you don't feel it happening, you just notice years later that the air band is gone and a ringing moved in that never leaves.

This is why Chapter 8 told you to calibrate a reference monitoring level and mix at it — a conversational, around-70-something-dB SPL level, not a loud one. Quiet mixing isn't just safer; it's more accurate, because the equal-loudness curves (Chapter 4) flatten out at moderate levels, so what you hear is closer to what you've actually got. Loud mixing flatters everything and fatigues your ears inside an hour, after which you can't trust a single decision you make. Mix quiet, take breaks, and when you go to a loud show or track a loud source, wear hearing protection — musician's earplugs lower the level without wrecking the frequency balance. Protecting your ears isn't caution. It's protecting your only irreplaceable instrument and the entire career you're building on it.

🔄 Check Your Understanding

1. You mix for three hours at a level that feels "exciting" — call it around 95 dB SPL. Roughly how does your safe-exposure time compare to mixing at 85 dB, and what happens to your decisions well before any damage shows up?
2. Why does the book keep telling you to set your monitoring level low and consistent rather than loud?

Verify

1. Every ~3 dB roughly halves the safe time, so ~95 dB cuts your budget to a small fraction of the ~8 hours that ~85 dB allows — you're into risk in well under an hour. Long before that, ear fatigue has crept in: bright sounds dull, your sense of balance drifts, and you start over-EQing to compensate. You can't trust a decision made on tired ears.
2. Two reasons. Accuracy: at moderate levels the equal-loudness curves are flattest, so your perceived balance is closest to the real one. Safety: low levels keep you far under the ~85 dB / 8-hour line and let you work all day without fatigue. A consistent level also means a mix that translates — you're always referencing the same loudness.

LUFS — Measuring Loudness the Way Ears Hear It

Peak meters answer "how close am I to clipping?" RMS meters answer "what's the average energy?" Neither answers the question that actually matters for a finished release: "how loud will this feel to a listener?" A delicate fingerpicked guitar and a wall of distorted synths can hit the exact same peak and the exact same RMS while feeling worlds apart in loudness, because your ear weights some frequencies far more heavily than others and cares about sustained energy more than instantaneous spikes.

LUFS (Loudness Units relative to Full Scale) is the measurement built to match perception, and it's why Chapter 33 retired peak-and-RMS for the loudness conversation. It does two things the old meters didn't. It applies K-weighting — a frequency filter that mimics your ear's sensitivity, discounting the deep lows you barely perceive and weighting the mids and highs you live in. And it gates out silence, so the quiet intro and the gaps between phrases don't drag down the reading the way a raw average would. The result is a number that tracks how loud something sounds, which is exactly what streaming platforms need to make different songs play back at an even level.

LUFS comes in three flavors, and you'll use all three:

Integrated LUFS is the headline number — the single loudness value for the entire track, and the one every platform compares against its reference to decide how much to turn you up or down.

Platform Loudness Targets

Here's the table the whole modern mastering chapter (Chapter 33) points back to. Read every number with one disclaimer stapled to it: these are the reference levels as of this writing, and platforms change them whenever they like. Don't memorize the digits; understand the mechanism.

The headline of Chapter 33 is the threshold concept worth repeating here: streaming normalization ended the loudness war. When the platform turns every track down to a common loudness, slamming your master to -7 LUFS no longer makes it louder than the next song — it just gets turned down to match, arriving with the same perceived loudness but less dynamic range and more distortion than the version you didn't crush. Loudness for its own sake buys you nothing on a normalized platform. Dynamics are a creative choice again. The right target is the one that serves the music: a punchy club-leaning electronic track might live around -9 to -11 LUFS and a dynamic singer-songwriter ballad might sit at -14 or quieter, and both are correct. The only universal rule left is the true-peak ceiling: keep your true peaks at or below -1 dBTP so nothing distorts on conversion or on cheap playback hardware.

True Peak vs. Sample Peak

That -1 dBTP ceiling deserves its own paragraph, because it catches people who think they're safe at exactly 0. Your DAW's standard peak meter reads sample peaks — the height of the actual samples stored in the file. But when a converter rebuilds those samples into a continuous analog wave for your speakers (Chapter 2), the smooth reconstructed curve can overshoot the highest sample, peaking between the sample points. That overshoot is an inter-sample peak, and it's why a file that reads exactly 0 dBFS on a sample meter can actually clip when it plays back.

True-peak metering (dBTP) predicts those between-sample overshoots by oversampling, so it tells you the real ceiling your file will hit on playback. Lossy encoding (the MP3 and AAC that streaming serves) can push inter-sample peaks even higher. That's the whole reason the recommended mastering ceiling is -1 dBTP instead of 0 — that 1 dB margin absorbs the overshoots and the encoding wiggle so your master stays clean on every device, not just the one you mastered on.

One Map: The Air, the File, the Listener

Pull it together. Three scales, three zeros, one signal passing through all of them — from a microphone in a room to a number a listener's brain assigns. Keeping them straight is what separates "my levels are a mess and I don't know why" from a session you control end to end:

  THE AIR              THE FILE                 THE LISTENER
 (acoustic)            (digital)                (perceived)
  ───────               ───────                  ───────
  dB SPL                dBFS                     LUFS
  0 = threshold         0 = clipping CEILING     measured vs. reference,
      of hearing            (audio is below)     ear-weighted (K) + gated
  ↑ louder = up         ↓ levels go DOWN         ↑ louder = up toward 0
  your ears (!)         your meters              the platform's meter

   sound  ──►  MIC  ──►  A/D  ──►  DAW  ──►  D/A  ──►  SPEAKER  ──►  ears
   (dB SPL)              (now dBFS, all negative)     (back to dB SPL)
                                  │
                          master measured here
                              in LUFS + dBTP
                                  │
                        platform normalizes to its
                         reference → consistent
                          loudness in the listener's ears

The three-zero map: dB SPL measures the air (0 at the bottom, your irreplaceable ears at the end), dBFS measures the file (0 at the top, everything below it), and LUFS measures what the listener finally perceives — the number platforms use to even everyone out.

Read left to right, that's the whole signal chain from Chapter 3 with the right scale labeled at every stage. A sound leaves a source at some dB SPL and hits the mic. The converter turns it into a file where every level is dBFS, negative, ceiling at zero — the world of Chapters 21 and 22, where you protect headroom and gain-stage. At the end you measure the finished master in LUFS and dBTP (Chapters 32 and 33), the platform normalizes it to its reference, and it lands back in the air as dB SPL in a listener's ears — the only meter that was ever the point.

Three zeros. Three references. Ask "compared to what?" every time, protect the headroom in the file and the hearing in your skull, and the decibel goes from the most confusing number in audio to the most useful one you own.