Mastering Tools and Techniques: EQ, Compression, Limiting, Stereo Enhancement, and Loudness

The Sticky Note

Five WAV files. Fourteen months of work, and it all fits in one folder: GLASS HOURS — MIX PRINTS.

Jaylen Cole printed them the way Chapter 31 taught him to print them — 24-bit, peaks breathing around -6 dBFS, not a limiter in sight on the 2-bus — and every one of the five survived its next-day listen. "Low Light." "Static Bloom." "Glovebox." "2 A.M." "Glass Hours," the title track. Each one got its single focused revision pass and then he made himself stop, because Chapter 31 also taught him that a mix you keep poking is a mix you're slowly un-finishing.

So the mixes are done. Done-done. And now there's a decision sitting on his desk with its arms crossed.

Demi sent him the write-up from Wren & Hollow's A/B day — the afternoon they sat in a mastering engineer's room and heard their EP come back polished by someone with full-range monitors and twenty years of finished records in their ears. Her verdict was the honest kind: "It's not different. It's finished. Like someone wiped the glass." Theo's addendum was practical: it cost real money, it was worth real money, and the engineer caught two things in their mixes that neither of them could hear in the spare bedroom.

Jaylen does the math anyway. Five tracks. Phone-repair-shop wages. The math does what math does.

But it's not only the money, and he'd tell you that if you asked. Wren & Hollow's record needed an outside finisher because it's going to live or die on subtlety — brushes, banjo, a voice that whispers. Jaylen's EP is going out into a world he's been studying since Chapter 1, on his timeline, at his budget, and there's a version of this where paying someone else to turn the last five knobs means never learning what the last five knobs do. He's read the honest list from Chapter 31 — when self-mastering is fine, when it's malpractice — and a five-track debut EP from a bedroom producer with trained ears, a treated room, and a reference playlist he's worn grooves into sits comfortably on the fine side of that list. Singles and demos and debut EPs: this is exactly the case this book teaches.

There's one more thing on his desk, and it's the most important object in this chapter. It's a sticky note, stuck under his monitor, and it says:

BETTER, OR JUST LOUDER?

He wrote it the night before, after a rehearsal run that almost went wrong. He'd loaded "Static Bloom" into a fresh session, dropped FL Studio's stock limiter on it — your DAW ships one too; Appendix E has its name — and pushed the gain up to see what would happen. What happened was glory. The track got dense, confident, radio. It sounded like money. He pushed it further and it sounded like more money. At 1 a.m., with the meters dancing and the track roaring, he reached for the keyboard to bounce it.

And the sticky-note thought arrived — early, for once. In Chapter 23 the morning had to catch him. This time he caught himself the same night: a limiter turns the whole track up. You are listening to LOUDER. You have not yet heard whether it's BETTER. He pulled the limited version down until it matched the unlimited mix — the matched-loudness law from Chapter 22, the one he now applies the way you apply a seatbelt — and A/B'd them honestly.

The glory evaporated. At matched loudness, his +6 dB of limiter push wasn't bigger. It was smaller. The kick had stopped landing. The hats had a fizz on them like the tops of the waveforms had been sanded. The 808's tail — the one he tuned in Chapter 13, saturated in Chapter 26, sidechained in Chapter 28 — had hardened into something that buzzed instead of bloomed.

Louder, yes. Better, no. The two are not the same thing, and at this stage of the pipeline — the last stage, the quarter-decibel world, where every move you make lands on every element at once — confusing them is the single most expensive mistake available to you.

So he wrote the sticky note, went to bed, and tonight he starts over, properly. Track one. The chain built knob by knob, in order, with reasons. A referee on the desk.

This chapter is that chain. By the end of it you'll have built one too — every stage with its job, its settings range, its budget, and the question it has to answer before it's allowed to stay. Five stages. None of them magic. All of them judgment.

The sticky note comes free.

🏃 Fast Track: Comfortable with everything Part V and VI taught — matched-loudness A/B, glue compression, M/S, gain-reduction stacking? Read the chain-flow diagram and "The Chain in Order" straight through (it's the spine), slow down for "The Limiter" and the true-peak sidebar, then hit "Limiting in Stages," "Sequencing an EP," and the Project Checkpoint. Do exercise C2 — the 0.5 dB ladder — tonight. That single drill is half this chapter.

🔬 Deep Dive: The full mastering literature is deep and worth the trip: Bob Katz's Mastering Audio is the standard text (Further Reading maps the route), and the psychoacoustics underneath every judgment call here — equal-loudness curves, why density reads as quality — lives in Chapter 4 of this book and in the companion volume, The Physics of Music, Chapter 5. The true-peak story leans on sampling theory from our Chapter 2; the companion volume's Chapter 32 goes deeper on reconstruction. This chapter gives you the working version: the chain, the budgets, the listening.

You Already Know What a Master Sounds Like

You've heard this chapter's subject your whole listening life. You've just never caught it working.

Put on a great album — front to back, any era, any genre — and notice what you don't do: you don't touch the volume knob. Track three is a wall of guitars and track four is one voice and a piano, and somehow the quiet song isn't quieter in the way that makes you reach for the remote; it's quieter in the way a room gets when someone starts telling the truth. Different songs, recorded months apart, mixed on different days in different moods — and they arrive as rooms in one house. Somebody did that. Somebody decided exactly how much quieter the quiet song should be, how long the silence lasts before it starts, and whether its top end should glow or hush. That's mastering — the sequencing-and-level craft of it, the part nobody notices because its whole job is to not be noticed.

Now think about the other half — the half you can hear. Recall the last time you heard a rough demo and then a finished record back to back. The demo might have a great song in it, a great mix even. But the record feels — pick your word — denser. Finished. Taller by an inch. Like the demo printed on paper and the record printed on card stock. Hold both at the same loudness and the difference is still there: it's not level, it's density — the quiet details riding closer to the loud ones, the whole thing carrying more information per second without shouting. That's the limiter and its colleagues, used well.

And you've heard them used badly, too. The track that's huge for thirty seconds and exhausting by ninety. The chorus that's somehow loud and small at the same time, pressing against glass. You learned that sound in Chapter 23 at the scale of one drum bus. This chapter it returns at the scale of the whole record — where the same mistake costs the whole record.

One craft, two faces: the invisible one that makes albums feel inevitable, and the audible one that makes singles feel finished. Both come out of the same short chain. Let's build it.

The Chain in Order: Five Stages, Every Knob Earned

First, the why of the order — because the order isn't a preset, it's a logic.

You learned the small version of this logic in Part V: EQ before compression when you want the compressor to react to the fixed tonal balance, not to the problem (Chapter 22). It scales up. In mastering, corrective moves come before dynamics moves because every dB of mud you remove is a dB the compressor and limiter never have to fight. Dynamics come before loudness because glue changes the peaks the limiter will see. And loudness comes dead last because the limiter is the only stage whose output is fixed against a ceiling — everything you do after a limiter either undoes it or clips. Each stage hands the next a cleaner problem. Run the chain out of order and each stage hands the next a fresh one.

Second, the stakes. In Chapter 22 a 3 dB cut was a scalpel move on one track among thirty-four. Here, every move lands on the entire record at once — the vocal, the kick, the air around the hats, the tail of the reverb, everything, together. That's why mastering is the quarter-dB world Chapter 31 promised: a 1 dB shelf at mastering does roughly the work a 3–4 dB move does in a mix, because nothing escapes it. Small hands. Wide brushes.

Third, the setup, before a single plugin loads:

• Fresh session, fresh day. Not the mix session — a new project, your printed 24-bit mix on one track. Mastering with the mix session open is how "one quick fader tweak" un-prints your mix. And not mix day: Chapter 31's next-day rule is doubled here, because mastering judgment is the first thing fatigue eats.
• Your references on a second track, routed around your processing chain so they stay untouched. These are the same three-or-more masters you built in Chapter 4 and have leaned on ever since.
• Your calibrated listening level (Chapter 8), your treated room (Chapter 10), and your headphones for cross-checks. You're about to make quarter-dB decisions; make them somewhere you've learned to trust — and verify everything later on the systems that don't flatter you.

Here's the whole chain at a glance. The budgets matter more than the tools: they're ceilings, not goals, and an empty stage is a legal stage.

THE SELF-MASTERING CHAIN  ·  signal flows top to bottom

   PRINTED MIX   24-bit WAV · peaks ≈ -6 dBFS · no 2-bus limiter   (Ch 31 spec)
       │
       ▼
  [0] GAIN-MATCH THE REFERENCES      refs pulled DOWN to the mix — never
       │                             the mix up · judged by average, not peak
       ▼
  [1] CORRECTIVE EQ                  budget: moves ≤ 2 dB, most ≤ 1 dB
       │                             wide Q · 1–4 bands · M/S option for
       │                             side-mud · linear-phase available
       ▼
  [2] GLUE COMPRESSION               budget: ≤ 2 dB GR, loudest moments only
       │                             1.5–2:1 · slow attack · auto release
       ▼
  [3] CHARACTER / SATURATION         budget: crumbs — if it's plainly audible
       │   (optional)                it's too much · often bypassed entirely
       ▼
  [4] LIMITING — IN STAGES           push in 0.5 dB steps, listening each step
       │     A "the leveler"  ~1–2 dB GR
       │     B "the wall"     ~1–2 dB GR · ceiling -1.0 dBTP · true-peak ON
       ▼
   THE MASTER  →  translation insurance (Ch 30 protocol)  →  print

The chain in order, with each stage's working budget — ceilings, not goals.

Stage 0 — Gain-Match the Reference: The First Move Is Humility

Before you process anything, you fix the scale you'll judge everything by. This is the reference-track compass from Chapter 4 doing its most important tour of duty, and the direction of the gain-match matters enormously.

Your references are finished masters. Their average level — the loudness your ear actually tracks, the VU-style average from Chapter 21, not the peaks — sits far above your printed mix, which is deliberately carrying -6 dB of peak headroom and no limiting. Play them back to back raw and the comparison is rigged before it starts: the ref will sound bigger, brighter, deeper, better, for reasons that are mostly just louder. You proved this to yourself in Chapter 22. At mastering it's not a bias, it's an ambush.

So: pull the references down to the mix. Put a trim or clip gain on the reference track and drop it — usually somewhere in the 6–10 dB range, it varies per ref — until you can switch between your mix and the reference mid-section and feel no level flinch. Match by ear on the average, on the body of the sound; an honest check is to flip between them while watching nothing, then confirm against an average-reading meter if you have one. Never push your mix up to meet the refs — your headroom is the limiter's working room later, and you don't spend it on a comparison.

Now listen. Switch between your mix and each reference, section against section, and write the gap list: ref has more air; my low-mids are thicker; ref's bass is tighter but mine is wider; my top is a touch dark. In Chapter 4's band vocabulary, walk sub, lows, low-mids, mids, high-mids, air, and score yourself against the playlist. That list — three to six honest lines — is the to-do list for stage 1. The gap is the to-do list; it has been since Chapter 4. You're just finally reading it at the final stage.

One caution that saves people: not every gap is a defect. If your reference is a club record and your track is a bedroom confession, part of the gap is identity — Chapter 18 taught you to keep deliberate violations. The gap list tells you where you differ; your manifesto from Chapter 5 tells you which differences were on purpose.

Stage 1 — Corrective EQ: Broad Strokes in the ≤2 dB World

Master-stage EQ is the same tool you mastered in Chapter 22 wearing very different gloves. In the mix you EQ'd one instrument at a time, surgically, sometimes savagely — a 6 dB cut on a boxy tom hurt nobody else. Here you're EQ'ing a finished balance. Boost 2 kHz and you boost the vocal's presence, yes — and the snare's ring, the guitar's edge, the hat's tick, and the part of the pad that was hiding politely under all of them. Every move is a committee decision now, which is exactly why the moves get small.

The working rules:

• Wide strokes. Q values around 0.5–1.0 — gentle hills, not notches. You're tilting balances, not removing resonances; resonance surgery was the mix's job (and if you find one now, Chapter 31's law applies: the mix is in the next room, go fix it there).
• Small moves. Most master EQ lives at 0.25–1 dB; 2 dB is the do I actually have a mix problem? line. If your gap list seems to demand more than 2 dB anywhere, that's not a mastering note, that's a mix note — reopen the mix, fix the source (Chapter 30's stage-of-origin discipline), reprint. Twenty minutes now beats a master that fights its own mix forever.
• Few bands. One to four moves. A typical self-master needs: one low-mid de-thickening, maybe one gentle shelf at the top, occasionally a sub decision. If you have seven bands working, you're mixing again.
• Matched-loudness A/B on every move. The Chapter 22 law, now at quarter-dB stakes. Any EQ boost makes the track louder; louder reads as better; you know this dance. Match and verify, every single time.

What do the moves usually look like? The same suspects you've been arresting since Chapter 22, at album scale:

The usual master-EQ moves. If any cell wants more than 2 dB, the mix is the patient.

That last row is your Chapter 28 investment paying its dividend. Mid-side EQ lets you treat the center of the stereo image and the edges separately — and at mastering this solves a problem ordinary stereo EQ can't: side-mud. Wide pads, doubled guitars, big reverbs — the wide, lush stuff — tend to deposit low-mid energy in the sides, where it adds thickness without adding power. A gentle high-pass on the side channel around 100–150 Hz, or a half-dB side cut at 200–300 Hz, cleans the edges of the image while the kick, bass, and vocal in the mid channel never feel a thing. The lows anchor to mono — Chapter 25's doctrine, now enforced at the final gate — and the record gets clearer and feels wider, because width you can't hear through is just fog.

And here, finally, is where you earn a term this book has been saving: linear-phase EQ. Every EQ you've used so far — every analog EQ ever built and most plugins by default — is minimum-phase: when it changes a band's level it also shifts that band's timing slightly. In a mix this is almost always irrelevant. But a linear-phase EQ does its filtering in a way that delays all frequencies equally — no phase shift between bands at all. Mastering reaches for it in two situations: M/S work, where keeping the mid and side channels phase-coherent protects your mono fold-down, and any move where you're EQ'ing material that will be summed or compared against an unprocessed version. The bill (Chapter 22's sidebar warned you): real latency, and pre-ringing — a faint smear that arrives before sharp transients, most audible as a softening on the front of heavy low-end hits at steep, surgical settings. At mastering's gentle, wide settings, the honest truth is that minimum- and linear-phase usually pass a blind test against each other; the working habit is linear-phase for M/S moves, either for the rest, and a quick ear on your kick and 808 attack whenever you choose linear — if the front edge of the kick got politely vague, you just met pre-ringing.

🔄 Check Your Understanding

1. Why do the references get pulled down to the mix's level, instead of the mix getting pushed up to theirs?
2. Your gap list says "low-mids way too thick — needs about -4 dB at 250 Hz." What does this chapter say you should actually do?
3. You make a side-channel EQ move in M/S mode. What check is mandatory immediately afterward, and which chapter's doctrine is it?

Verify

1. Two reasons: the comparison must happen at matched loudness or louder-sounds-better rigs every judgment — and your mix's -6 dB headroom is the limiter's working room later, not to be spent on a comparison. Refs come down; the mix never goes up.
2. Reopen the mix and fix it at the source — a move bigger than about 2 dB is a mix problem wearing a mastering costume (Chapter 31's "polish, not surgery"). Reprint, then re-master.
3. A mono check (and a glance at the correlation meter) — Chapter 25's mono-compatibility doctrine. M/S moves change what survives the fold-down, and at master stage a mono failure ships on every Bluetooth speaker at once.

Stage 2 — Glue Compression: Chapter 23 at Mastering Scale

You've already built this stage once. The drum-bus glue you dialed in Chapter 23 — 2:1, slow attack, auto release, a needle swaying 2–3 dB with the backbeat — turned eight strangers into a kit. Mastering glue is the same move at the scale of the whole record, and at half the dose: a glue/bus compressor across everything, set so gently that its entire output is a feeling.

The settings barely move from track to track, which tells you something — this stage is a posture, not a fix:

• Ratio: 1.5:1 to 2:1. The gentlest slopes on the knob. You're leaning on the record, not gripping it.
• Attack: slow — 30 ms or more. The transients are off the table, permanently. You spent twenty-two chapters protecting punch; this is not where it dies.
• Release: auto, or slow enough to breathe with the song's phrases, not its hits. The bus-loving auto-release circuits from Chapter 23 were built for exactly this seat.
• Threshold: high. The GR meter should read zero through the verses and touch 1–2 dB at the loudest chorus moments — and come all the way home between them. If the needle lives below zero, you're mixing with a compressor again.
• Makeup: matched, so the bypass A/B is honest. Always.

What does 1–2 dB of whole-record glue buy? Sections that hold hands. The chorus still lifts — your Chapter 27 automation did that on purpose — but it sits down into the track as it arrives instead of stepping on it. Transitions feel inevitable. The record gains a little density, the sense that everything is happening in one room on one night.

And the cost, which is why the budget is hard: a master compressor is an automation eraser. Your +0.5 dB chorus lift, your phrase-by-phrase vocal rides at ±1–2 dB — a 2:1 ratio working 4 dB deep will quietly halve them. You'd be paying a machine to undo the most painstaking pass of your mix. That's the why behind ≤2 dB of gain reduction: below that line the glue compresses the moments between dynamics; above it, it starts compressing the dynamics themselves.

When do you skip this stage entirely? Often. If your mix already wore a bus compressor (plenty do — yours may have, from the Chapter 26 color pass or a 2-bus glue experiment), the record arrived glued, and a second helping is a tax on a paid bill. Bypass the stage, A/B at matched loudness, and let the better version win without checking whose feelings it hurts. An empty insert is not a missing skill. It's a verdict.

Stage 3 — Character: Crumbs, Optional

Chapter 26 taught you the whole flavor cabinet — tape's soft squeeze, tube's even warmth, console edge — and the drive-and-match workflow that keeps it honest. At mastering, the cabinet reopens for one last, tiny serving: a half-dB of drive on the whole record can add density and the kind of harmonic glow that helps small speakers find the bass (the phone-speaker salvation from Chapter 26, applied globally).

The dose rule here is the strictest in the book: homeopathic. If a listener — or you, tomorrow — can point at the saturation, it's too much. You're seasoning a finished dish someone else will eat in one bite.

The honest workflow: engage the saturator at its gentlest setting, drive until you can barely tell, pull the output down to match (drive-and-match, Chapter 26), then A/B at matched loudness and ask the only question: better, or just different-and-louder-flavored? Watch the low-mids especially — saturation's harmonics pile into the 200–400 Hz neighborhood you may have just spent stage 1 cleaning. Nothing is more common than a character stage that politely re-congests the exact mud the EQ stage removed. That's not a paradox; that's the tradeoff ledger doing its job.

And hear this clearly: most well-colored mixes need none. If you did Chapter 26's color pass — vocal tape, bass tube, drum-bus console, crumbs on the 2-bus — the character is already printed in the mix, where it belongs, applied per-element where you could control it. A master-stage saturator earning its bypass is the chain working, not failing. The amateur adds; the professional subtracts; the mastering professional mostly declines.

🧩 Productive Struggle: Before you read a word about the limiter, go meet it with your own ears — this is the chapter's most important ten minutes. Open your printed mix in a fresh session. Drop your DAW's stock limiter on it. Set its ceiling (the output-level control) to -1 dB. Now raise the limiter's input gain 1 dB at a time, listening hard at each step to three things: the front edge of your drum hits, the tail of your bass or 808, and the busiest bar of your loudest chorus. Somewhere up that ladder, something will complain — write down the gain value where you first heard it and what you heard, in your own words. Then come back. You've just located the edge this entire section is about, on your own material, before anyone told you where it should be.

Stage 4 — The Limiter: Where Better-or-Louder Gets Decided

Everything before this was preparation. This is the loudness stage — the one move in the chain that exists to change how loud your record is, and therefore the place where the sticky note earns its keep.

You met the limiter in Chapter 23 as a wall: effectively ∞:1 compression, near-instant attack, usually with lookahead, letting nothing past its ceiling. Its mastering job is the inverse of how it's drawn: a limiter doesn't push peaks down so much as it raises the floor toward the ceiling. You add input gain; the whole record rises; the peaks that would cross the ceiling get caught and shaved; everything else — the body, the quiet detail, the average level your ear reads as loudness — comes up underneath. Density, manufactured.

Two terms, both of which this chapter owns and you'll use forever:

The ceiling is the output level your limiter will not let a peak cross — the highest point of your finished master. Set it to -1.0 dBTP, with true-peak mode switched on, and leave it there. Not 0 dBFS. Not -0.1. The sidebar below explains exactly why the extra margin exists, and the short version is: your master will be played back through reconstruction filters and lossy encoders that swing between your samples, and -1 dBTP is the standard margin — recommended by the platforms themselves, as of this writing — that keeps those swings from clipping gear and encoders you'll never see. The full measurement story, and the loudness units that go with it, is Chapter 33's; today the ceiling is a setting, and the setting is -1.0 dBTP.

The push is the input gain you add — how hard you drive the record into the ceiling. The push is the decision. Here's the ritual:

Push in 0.5 dB steps. Not to a number from a forum. Not to where the meter looks impressive. Half a dB at a time, and at every step, run the loop:

THE BETTER-OR-LOUDER LOOP  ·  run at every +0.5 dB of limiter push

        ┌──────────────────────────────┐
        │      push +0.5 dB more       │◄────────────┐
        └──────────────┬───────────────┘             │
                       ▼                             │
        LISTEN, at each step, to:                    │
          · transient edges (snare crack, hat tick)  │
          · bass / 808 tails (grind? hardening?)     │
          · the densest bar of the loudest section   │
          · your own attention (fatigue is data)     │
                       ▼                             │
        match loudness · A/B against the last step   │
                       ▼                             │
            "better, or just louder?"                │
              │                   │                  │
           BETTER             JUST LOUDER            │
        (denser, still        (or actively worse:    │
         punching, no          splash, grind,        │
         complaints) ─────►    glass)                │
              keep climbing       │                  │
              ──────────────────► ▼                  │
                          STEP BACK 0.5 dB ──────────┘
                          that's your master

The decision loop. The master is the last step that answered "better" — not the loudest step you survived.

What you're listening for at each rung is the distortion-vs-density listen — the audible signature of a limiter starting to lose:

• Transients dull first. The snare's crack rounds off; hats and cymbals develop a splashy fizz, like the top of the waveform got sanded. Dense cymbal-heavy material shows this earliest.
• Low end hardens. Long bass and 808 tails are the limiter's hardest problem — a sustained low note is the level, so gain riding on it becomes audible grind and buzz. Trap and bass-heavy electronic material breaks here first.
• The record gets smaller as it gets louder. The Chapter 23 paradox at full scale: past the edge, density stops reading as confidence and starts reading as pressure. Glass. The mix stops breathing.
• You get tired. Honestly the most reliable meter you own. If looping the chorus is becoming unpleasant, your limiter found the edge a few rungs ago.

Where's the edge in numbers? It depends on the mix, the genre, and the arrangement density — a sparse piano ballad shows limiting at +2 dB of push; a wall-of-sound chorus might take +6 happily. Typical self-masters of well-made mixes land somewhere in the +3 to +6 dB range of total push, and you should treat even that range with suspicion: it's your ears' call, made through the loop, one honest half-dB at a time. Where the target loudness should land for streaming platforms — the actual numbers, the meters that read them, and the genuinely good news about why this war is over — is the entire business of Chapter 33. Today you set loudness the way every number should have been set in the first place: by ear, against references, at matched loudness.

One trap to disarm before you run the loop: the limiter's own bypass lies. Bypassing a limiter drops the level by the entire push — so engaged-vs-bypassed always flatters engaged, massively. Some limiters ship a gain-matched listen mode (sometimes labeled with words like "delta" or "auto-match" — Appendix E); use it if you have it. If not, do it the durable way: bounce candidates and compare them as files, levels matched by hand. Slower. Honest. Worth it.

🔍 Why Does This Work? Why does louder masquerade as better so reliably — even to ears as trained as yours are by now? Two mechanisms, stacked. First, the equal-loudness curves from Chapter 4: your ear's frequency response flattens as level rises, so the same material played 1 dB hotter genuinely delivers more perceived bass and treble — it isn't an illusion, it's free EQ in the flattering direction, and it arrives with every unmatched A/B. Second, density itself carries meaning: decades of radio, club systems, and finished records have taught every listener that dense equals professional, so a limited track borrows that association even when the limiting has begun eating the punch that actually made the record exciting. Matched-loudness comparison surgically removes the first mechanism and starves the second — which is why the law you learned at one EQ move in Chapter 22 becomes load-bearing at mastering, where every single stage changes level as a side effect. The loop isn't a ritual of caution. It's the only experimental design that isolates the variable you care about.

⚙️ Advanced Sidebar: True Peak vs Sample Peak — Why the Ceiling Is -1 dBTP

Your DAW's peak meter has been telling you a small lie since Chapter 2 — harmless until today.

A digital file is dots: discrete samples, 48,000 of them per second in your session. Your peak meter reports the highest dot. But you learned in Chapter 2's threshold concept that the audio isn't the dots — playback reconstructs the one smooth, band-limited wave that passes through all of them, and that wave is under no obligation to peak exactly at a sample. Between two high samples on a curve's shoulder, the reconstructed wave bulges higher than either:

0 dBFS ─────────────────────────────────────  the digital ceiling
                    ⚠ true peak ≈ +0.4 dBTP
                   ___
            ●    ╱     ╲    ●                 ● = samples (all below 0 dBFS —
          ╱    ●         ●    ╲                   sample-peak meter reads -0.2
        ╱                       ╲                 and reports "no clipping")
      ●                           ●           ─── = the reconstructed wave,
                                                  which swings BETWEEN samples

An inter-sample over: every sample is legal, and the actual wave clips anyway.

That bulge is an inter-sample over (often written 0 dBFS+). A master whose samples all sit at -0.2 dBFS can carry a reconstructed waveform that exceeds 0 — and the places it gets reconstructed are exactly the places you can't supervise: the D/A converter in a cheap Bluetooth speaker, and — worse and more common — the lossy encoder at every streaming platform. MP3 and AAC encoding redraws the waveform slightly, and on hot material those redrawings overshoot: a file that measured "legal" goes in, and a file with crackly, spitty clipping on the transients comes out, with your name on it.

True-peak metering exists to catch this: it oversamples the signal — recalculating it on a much finer time grid, typically 4× or more — to estimate where the reconstructed wave actually lands, and reports the result in dBTP (decibels true peak). The measurement is standardized in ITU-R BS.1770, the same document Chapter 33's loudness units come from.

So the ceiling practice writes itself: set the limiter's ceiling to -1.0 dBTP with true-peak detection on, and the whole problem evaporates — a full decibel of margin absorbs reconstruction bulges and encoder overshoot both, and as of this writing -1 dBTP is the ceiling the major platforms themselves recommend. The old habit of -0.1 or -0.3 sample peak dates from CD masters and meters that couldn't see between the dots. You can. Spend the decibel; it's the cheapest insurance in this book.

🔄 Check Your Understanding

1. What is the ceiling, where does this chapter set it, and what two downstream systems does the margin protect?
2. Describe the limiter push ritual: the step size, the three places you listen at each step, and what decides where you stop.
3. Why does a quick engaged-vs-bypassed check on a limiter always flatter "engaged" — and what's the honest alternative?

Verify

1. The ceiling is the output level the limiter lets nothing cross — the master's highest point. Set it at -1.0 dBTP, true-peak mode on. The margin protects D/A reconstruction in playback gear and, especially, lossy encoders (MP3/AAC) whose redrawn waveforms overshoot on hot material.
2. Push input gain in 0.5 dB steps. At each step listen to transient edges (snare/hats), bass and 808 tails, and the densest bar of the loudest section — plus your own fatigue. A/B against the previous step at matched loudness and ask "better, or just louder?" The master is the last step that answered better; when the answer turns, step back 0.5 dB.
3. Bypassing removes the entire push, so the comparison is loudness-rigged by definition — engaged is louder, louder reads as better. The honest alternative: a gain-matched listen mode if the limiter has one, or bouncing candidates and comparing files level-matched by hand.

Before the judgment calls, here's the whole chain again as a settings card — the thing to tape inside your notebook next to the sticky note:

The chain as a settings card. Budgets are ceilings; bypass is always a legal answer.

The Judgment Calls: The Tools That Ask Questions First

The chain above will master most well-made mixes. The tools below are the ones that come with interview questions — each is genuinely useful, each is on the box of every "mastering suite" ever sold, and each does more damage in confident hands than honest ones.

Multiband Compression, Honestly

A multiband compressor splits the record into frequency bands at crossover points — say lows below 120 Hz, low-mids to 800 Hz, highs above — and gives each band its own compressor: its own threshold, ratio, attack, release. On paper it's the ultimate mastering tool: dynamics control, per neighborhood.

Chapter 28 flagged this one with a caution and promised the honest conversation here. Here it is: a multiband compressor is a fix-a-mix-problem tool, and you own the mix. Look at what it's actually for: the bass blooms out of control only in the choruses (a level/automation problem); the vocal's sibilance survived into the master (a de-essing problem — Chapter 29); the low-mids swell when everything plays at once (an arrangement-density problem — Chapter 16, wearing its Chapter 20 disguise). Every one of those has a stage-of-origin fix that's cleaner, more surgical, and free of collateral. A professional mastering engineer reaches for multiband because they can't open the mix — it's their scalpel-through-the-keyhole. You are not through the keyhole. The session is in the next room. The multiband moment is usually the go-fix-the-mix moment, and recognizing it is a skill this book has been building in you since Chapter 30.

The honest residual uses, once you've asked the question and the mix really is closed (someone else's project, a deadline, source files gone): gentle high-band compression as a whole-mix de-esser; one band leashing a bass that misbehaves only sometimes, 1–2 dB, slow. And the costs even then: crossover filters smear phase and blur the seams between bands, and any band compressed independently detaches tonally from the record — the "rubber bass under a still photograph" sound of preset-driven multiband mastering. Which is the last warning: mastering-suite presets ship with four bands armed and working. Bypass is a setting. Start there.

A gentler cousin usually beats it anyway: dynamic EQ (Chapter 28) — an EQ band that ducks only when its frequency range misbehaves, no crossovers, no seams. If a real master-stage frequency-dynamics problem survives the go-fix-the-mix question, try one dynamic band at the trouble spot before you un-holster four crossovers.

Stereo Enhancement: The Caution Section

Every mastering suite has a width knob, and the width knob has the same problem as the loudness knob: wider reads as better on first listen, the way louder does. More spread, more "more." It's loudness's cousin, and it demands the same court procedure.

What master-stage width tools actually do, ranked by honesty:

• Side-level nudges — the honest one. In M/S terms (Chapter 28): turn the side channel up or down a touch. ±0.5 to 1 dB is the working range; ±1 dB is the whole budget. A genuine half-dB of side lift can open a mix that arrived a little narrow. It cannot create width that the mix never had — width is made of arrangement and panning and doubles (Chapters 16, 25), not of a knob.
• Frequency-dependent widening — defensible when it's really the Chapter 25 doctrine wearing a new badge: keep lows mono (the side high-pass from stage 1), let the highs carry the width.
• Haas-style delay wideners — Chapter 17 already sent you the bill on these: timing tricks that collapse into comb filtering when summed. On one synth in a mix, a managed risk. On a finished master, where the fold-down is the record on every phone and Bluetooth speaker — no. The bill arrives everywhere at once.

The law, regardless of tool: mono-check after every single width move. Not at the end of the session — after every move, with the correlation meter from Chapter 25 in view (positive and comfortable; flirtations with zero mean trouble). At mix stage, a width mistake cost you one element. At master stage it costs the record, on every mono-summing system on earth, simultaneously. Chapter 25's doctrine was a habit. Here it's law.

And ask the prior question first: if the master needs widening, why is the mix narrow? The honest answer is usually a pan map or arrangement note — and the session is right there.

Limiting in Stages: Two Gentle Beat One Slammed

Chapter 28 gave you the gain-reduction stacking doctrine — 2 dB from each of three stages sounds cleaner than 6 dB from one, because each stage works inside its own transparent zone. At the final gate, the doctrine gets its most valuable application: limiter stages.

Instead of one limiter doing all the pushing, run two in series, each doing roughly half the work:

ONE SLAMMED                          TWO GENTLE  (same total density)

mix ──► [ LIMITER  −5 dB GR ]        mix ──► [ A: leveler  ~2 dB GR ]
            │                                      │  slower release —
            │ one circuit panicking                │  rides the phrases
            ▼                                      ▼
        splashy hats · hard 808              [ B: the wall  ~2 dB GR ]
        tails · "glass" by bar 8                   │  fast · true-peak ·
        fatigue by chorus two                      │  ceiling −1.0 dBTP
                                                   ▼
                                             same loudness — punch intact

Stacking at the final gate: each stage stays inside its transparent zone.

The division of labor: stage A is the leveler — slightly slower release, doing 1–2 dB of broader, musical gain-riding on the loudest phrases; it hands stage B a track whose peaks are already polite. Stage B is the wall — the final ceiling at -1.0 dBTP, true-peak on, faster and more surgical, catching 1–2 dB of what's left. Each stage stays under its own audibility threshold; their small artifacts don't gang up the way one circuit's panic does. Same total push, same density — and the snare still cracks.

(You'll also meet engineers who use a soft clipper as stage A, shaving transient tips with Chapter 26's saturation math before the limiter ever works. It's a real technique with a real sound — a little attitude on the edges — and the same rules apply: tiny doses, matched A/B, your ears as the court. Test it; don't cargo-cult it.)

The push ritual doesn't change — 0.5 dB steps, the loop at every rung — you're just distributing the steps across two stages, usually alternating. And the verdict procedure stays the same too: bounce the staged version and a single-limiter version at the same loudness and let your next-day ears choose. They will not hesitate.

🔄 Check Your Understanding

1. The multiband moment on your own project is usually a signal to do what?
2. What's the full budget for side-level enhancement, and what must follow every width move?
3. In two-stage limiting, what does stage A do that helps stage B — and what doctrine from Chapter 28 is this applying?

Verify

1. Go fix the mix. Multiband exists to fix mix problems through a closed door — your door is open. Find the stage of origin (arrangement, source, edit, mix) and solve it there; reprint.
2. About ±1 dB of side level, total — and a mono check (with the correlation meter in view) after every single move, because a master-stage width failure ships on every mono-summing system at once.
3. Stage A (the leveler, slower) rides the loudest phrases 1–2 dB so the peaks arriving at stage B (the wall) are already polite — each stage stays inside its transparent zone. It's the gain-reduction stacking doctrine: several gentle stages beat one violent one.

The Reference-Mastering Workflow: Match Character, Not Just Level

Everything so far compares your master to references at matched loudness. The full workflow goes one layer deeper: you're not matching their level — normalization will re-deal those cards anyway, as Chapter 33 will demonstrate — you're matching their character, on two axes.

Tonal character. With refs gain-matched, walk Chapter 4's bands one at a time — sub, lows, low-mids, mids, high-mids, air — and score your master against each ref: darker or brighter, thicker or leaner, per band. Write it as a scorecard, not a feeling. Three refs are better than one precisely because they triangulate: if your track is darker than all three in the air band, that's information; if it's darker than one and brighter than another, that's taste space, and your Chapter 5 manifesto holds the tiebreaker.

Loudness character — which is not level. At matched loudness, how dense do the refs feel? Do their choruses lift and breathe, or press forward in one confident slab? Do drum hits poke through the wall or sit inside it? This is dynamics-as-aesthetic, and genres disagree about it on purpose. Your job is to know what your genre's records feel like at equal loudness and decide — deliberately, manifesto in hand — where yours sits on that spectrum. A folk record mastered to slab density is a translation error in the cultural sense: it arrives wearing the wrong clothes.

Then the final exam, the cheapest and most brutal test in mastering: the playlist test. Drop your bounced master into a playlist with your three refs. Shuffle. Walk away — do dishes, take the bus, live your life. When your track arrives unannounced among records you love, does it announce itself as the amateur in the room? Tonally? In density? In confidence? Every flinch is a note for the revision pass. When it stops flinching — when your track comes on and you just listen to it like it belongs — you're done. That's the whole goal of this chapter in one sentence: a record that sounds like it was invited.

Sequencing an EP: The Forgotten Half of the Craft

If you're releasing one single, skip ahead. If you're releasing an EP or album, welcome to the half of mastering nobody puts on the plugin box: deciding how five songs become one record.

Track order is an energy argument. You built energy curves inside songs in Chapter 16; an EP has one too, and the same physics: openers make promises, peaks need approaches, and contrast is the engine. The classic shapes — strong opener, single early (slot one or two), the quiet "sit-down" track guarding the middle, the heaviest hitter late, a closer that resolves — exist because they map how attention actually moves across eighteen minutes. They're conventions, not laws (Chapter 18's contract logic applies; break them knowingly). The classic mistake is front-loading: your two biggest tracks back to back at the top, and the EP spends its last ten minutes apologizing.

Test the seams, not the songs. You already know the songs work. What you haven't heard is the eight seconds where one ends and the next begins — and that's where sequencing lives or dies. Build a seam reel: the last twenty-odd seconds of each track, the gap, the first twenty of the next. Loop each seam. You're listening for three things:

• Key adjacency. Sustained tails — pads, 808s, reverbs — ring across gaps, and an ending chord that clashes with the next opening reads as a stumble even to listeners who can't name a single note. Adjacent and related keys flow; distant ones can work if the gap gives the ear time to release the old tonal center. (DJs take this seriously enough to have built entire key-wheel notation systems for set sequencing; you only need to audition your four seams.)
• Tempo adjacency. Tempo steps are gear changes. Small steps feel like momentum; a hard drop from your fastest track into your slowest feels like a stall — unless the gap is shaped to make it feel like a breath instead.
• The gap itself. Two seconds is the standard convention between tracks — and it's a default, not a decision. Shorten toward one second (or crossfade — the Beatles famously welded Abbey Road's side-two medley into one continuous build) when momentum should carry through. Lengthen toward three or four when the ear needs to exhale: after the emotional track, before the heavy one. Silence is a riser — the held breath before a drop is arrangement at EP scale.

Level relationships: the EP is the unit. Here is the mistake that ruins more self-mastered EPs than any plugin: mastering each track in isolation, each to its own maximum, so the quiet song arrives as loud as the bangers. A record's quiet track must be allowed to be quieter — that's its job; that contrast is the whole reason it works (you've known this since Chapter 16, and your favorite albums prove it every time the volume knob stays untouched). Master the tracks relative to each other: anchor the loudest track first, then place every other track against it by ear, in the sequence, at matched listening level. Expect your gentlest track to sit 1 to 2 dB under your loudest — and expect that to be exactly why it lands. Sparse arrangements expose limiting soonest (fewer elements, nowhere for artifacts to hide), so the quiet song usually wants the least push of the set, not the same push turned down.

EP LEVEL & ENERGY MAP  ·  Glass Hours  (level relative to loudest track)

   0 dB ┤                                  ████ "2 A.M."
        │            ████ "Static                        ████ "Glass
 -0.5   ┤            Bloom"                              Hours"
        │  ████
 -1.0   ┤  "Low
        │  Light"
 -1.5   ┤                       ████ "Glovebox"
        └────────────┬──────────────┬──────────────┬──────────────
          gap: 2.0 s │    3.0 s     │    3.0 s     │    2.5 s
          (momentum) │  (exhale)    │ (held breath │  (reset before
                     │              │  before the  │   resolution)
                     │              │  big one)    │

The Glass Hours level map: five tracks, one record. The quiet one sits low on purpose.

Tonal consistency is the other half of relative mastering: the tracks should sound like rooms in one house. Mixed across months, they won't, quite — one will be darker, one toppier. Small per-track EQ moves (the stage-1 toolkit, even smaller) bring the family resemblance; your master chain's shared stages do the rest. And finish with a tops-and-tails pass: trim each track's head close to the first sound, fade each tail down into silence rather than letting it fall off a cliff — the unglamorous thirty minutes that separates a record from a folder of bounces.

Case study 2 walks the entire Glass Hours sequencing session — the draft order that failed, the seam reel, and the one track that had to go back to the bench.

🔄 Check Your Understanding

1. What's "the unit" when you master an EP, and what's the canonical mistake that violates it?
2. Name the three things you audition at a seam.
3. Why does the quiet track usually want the least limiter push — not the same push at a lower fader?

Verify

1. The EP itself is the unit. The canonical mistake: mastering each track in isolation to its own maximum loudness, which steals the quiet track's contrast and makes the set read as one flat slab.
2. Key adjacency (do tails and openings clash?), tempo adjacency (does the gear change feel like momentum or a stall?), and the gap itself (default 2 s; shorter for momentum, longer for breath).
3. Sparse arrangements expose limiting artifacts soonest — fewer elements, nowhere to hide. The same push that's transparent on a dense banger is audible on the ballad, so the quiet track needs gentler limiting, not just a lower level.

Format Deliverables: What You Actually Print

The streaming era simplified deliverables wonderfully. Here's the honest inventory:

The streaming master — the one deliverable everyone needs: a 24-bit WAV at your session's sample rate (44.1 or 48 kHz both fine), peaks at or under -1.0 dBTP. As of this writing, every major distributor accepts exactly this and delivers it everywhere; check yours, but don't expect surprises. Don't convert sample rates unless a target demands it — and if one does, let a good resampler do it once, at the end. Name the file like Chapter 19 taught you (Artist_Track_Master_v1_2448.wav beats FINAL_master_FINAL2.wav in every timeline where you ever open this folder again).

16-bit versions exist for one reason: targets that demand them, chiefly CD — the Redbook standard is 44.1 kHz/16-bit, fixed since 1982. The rule from Chapter 2 stands: when you reduce bit depth, dither, once, at the very end of the chain — never earlier, never twice. (DDP, Chapter 31's vocabulary word, is the file format CD plants take; if you're pressing discs, your duplicator's spec sheet drives.)

Alternate masters — instrumental, clean/radio edit, a cappella if you'll want remix bait (Chapter 37 will) — get printed now, through the identical chain and settings, while everything is loaded and verified. Future-you, asked for an instrumental by a sync supervisor on a Tuesday deadline, will light a candle for present-you. Chapter 35 handles where these go.

Vinyl, in one honest section. Vinyl is the one format that gets a vote on your master, because vinyl is physics, not files. A cutting stylus carving a groove imposes real vetoes: heavy low end in the sides of the stereo image makes grooves the playback needle can literally jump out of — which is why lows-to-mono (Chapter 25) stops being a doctrine and becomes a cutting engineer's requirement; hot sibilance and dense top end distort toward the inner grooves, where the vinyl passes the needle more slowly and the wiggles crowd together; and loudness costs playing time — louder cuts need wider groove spacing, so a hot 24-minute side is a contradiction the lathe will not negotiate (somewhere under twenty-ish minutes a side is the commonly repeated comfort zone — treat it as a rule of thumb, not gospel). The honest advice in one line: if you're cutting vinyl, send the cutting engineer your dynamic master — not the streaming-loud one — and then actually talk to them. It's the one place in modern release life where a human with a lathe answers your file, and they know their machine.

Metadata — ISRCs, credits, the embedded who-owns-what — rides with distribution, not with the WAV you print tonight; Chapter 35 walks every field. Tonight your job is one clean, truthfully named, -1.0 dBTP master per track, plus alts.

Translation Insurance: Chapter 30's Protocol, One Last Time

You ran the translation gauntlet on your mix in Chapter 30. The master gets the same protocol — same systems, new questions, because limiting changes what each test is checking:

• The car — low-end balance and level in context: put the master in a playlist with your refs, shuffle, drive. Does it sit, or lurch? (For Jaylen this test carries some history. It goes fine. It keeps going fine. That's the point of the whole book.)
• The phone speaker — does the low end still imply? Your Chapter 26 harmonics carry the 808 here; the limiter shouldn't have hardened them into fuzz.
• Earbuds — the post-limiter sibilance check. Limiting raises average level, which pushes esses and cymbal fizz forward; if the earbuds flinch, revisit the push (or the mix's de-essing, Chapter 29, if it was marginal all along).
• Quiet listen — does the vocal still lead at whisper volume? Density should have helped this (the expensive-at-low-volume signature from Chapter 23); if quiet playback got worse, the limiter ate your foreground/background relationships.
• Next room — the balance-truth test, unchanged.

Plus the master-only test you already know: the playlist test, shuffled among references, which is translation insurance and reference workflow in one move. And the calendar rule that governs everything in Part VII: print, sleep, verify with tomorrow's ears, and only then call it done. A master discovered to be 0.5 dB too bright after release is a permanent lesson in why this paragraph exists.

Common Mistakes and the 60-Second Fixes

The greatest hits of self-mastering, with the diagnosis-to-fix turnaround this book owes you:

Ten classics. Every one of them is cheaper to fix tonight than to live with forever.

🔄 Check Your Understanding

1. What's the complete spec of the streaming master you print tonight?
2. Name vinyl's two physical vetoes and the one-line honest advice for vinyl releases.
3. Your master passes everything but earbuds, where the esses sting. Name the two candidate fixes and how you'd choose.

Verify

1. A 24-bit WAV at the session's sample rate (44.1 or 48 kHz), peaks ≤ -1.0 dBTP, truthfully named — plus alternate versions (instrumental/clean) printed through the identical chain.
2. Lows in the stereo sides make grooves needles jump out of (lows must anchor to mono), and hot top end/sibilance distorts at the inner grooves — plus loudness costs playing time. The advice: send the cutting engineer the dynamic master and talk to them.
3. Reduce the limiter push (limiting raised average level and pushed the esses forward) or revisit mix-stage de-essing (Chapter 29) if it was marginal all along. Choose by scale: if esses sting only at the final +0.5–1 dB of push, it's the limiter; if they're hot even at gentle push, the mix was the patient.

Project Checkpoint: Building "Static Bloom"

Where the track stands: "Static Bloom" went through the full chain this week, and case study 1 logs every knob. The shape of it: references trimmed down 9-ish dB to the printed mix; three corrective EQ moves, none over 1 dB (a -1 dB wide bell at 210 Hz where the chorus pad, guitar, and 808 overlap politely thickened; +0.75 dB of 10 kHz air against both refs; a side high-pass at 110 Hz that cleaned the image's edges — mono-checked, twice); glue at 2:1 touching 1–1.5 dB only when the double-chorus leans in; the saturation stage auditioned and bypassed, because Chapter 26's color pass already printed that flavor where it belonged; and the limiter built in two stages — a leveler doing not quite 2 dB and a true-peak wall at -1.0 dBTP doing the rest — pushed up the 0.5 dB ladder to +3.5 total, where the record got denser and kept punching. The +4 rung splashed the hats and hardened the 808's tail. The +5 print ("the loud one") lost the next-day blind at matched loudness, three listens out of three. The sticky note refereed all of it.

Your assignment — master your track:

1. Tomorrow, not today. Fresh session, printed mix (Chapter 31's spec) on one track, three refs routed around the chain.
2. Stage 0: refs down to the mix; write the gap list in band vocabulary.
3. Stage 1: corrective EQ from the gap list — wide, ≤2 dB, M/S side high-pass if the sides carry mud. Matched A/B per move; mono check after any M/S move. Any move that wants more than 2 dB sends you back to the mix, and that's the system working.
4. Stage 2: glue at 1.5–2:1, slow attack, auto release, ≤2 dB GR on the loudest moments only. Bypass test at matched loudness; keep the winner.
5. Stage 3: character crumbs if — and only if — the matched A/B says better, not just warmer-and-louder.
6. Stage 4: two limiter stages, ceiling -1.0 dBTP, true-peak on. Climb in 0.5 dB steps running the better-or-louder loop; stop at the last better; print that and one deliberately hotter version for tomorrow's blind verdict.
7. Translation insurance: car, phone, earbuds, quiet, next room, playlist-shuffle among refs. Then the next-day listen, then print the keeper: 24-bit WAV, -1.0 dBTP, named like an adult. Save the session — it's the template for every master you'll ever make, and Chapter 38 is going to want it for an experiment.

Genre adaptations. Hip-hop/trap/bass music: the 808's tail is the limiter's first casualty — run the loop's bass-listen on your longest sustained 808 note, and expect your genre's density culture to tempt you up the ladder (Chapter 33 has news about what that buys). Rock/band: cymbals splash first — listen there; glue earns more on a band mix if your 2-bus ran clean, and your genre's drama lives in dynamics, so guard the quiet-verse-to-big-chorus lift like it's the chorus itself. Electronic/dance: club density is legitimate — push further, but stage the limiting and keep sub width mono-anchored; the side high-pass is your friend forever. Folk/acoustic/singer-songwriter: sparse arrangements expose limiting soonest — yours is likely a +2-to-+3 record, and that is a feature; the playlist test against your genre's refs is the verdict that matters, not the loudness of someone else's. Spoken word: your "master" is mostly consistency and loudness-to-spec, and the numbers that govern it arrive in Chapter 33 — bring your Chapter 29 chain and wait one chapter.

Cross-Chapter Connections

Backward: This chapter is Part V and VI cashing out. The -6 dB of headroom your master chain just spent was banked in Chapter 21. The matched-loudness law from Chapter 22 became the better-or-louder loop. Chapter 23's glue and limiter-intro grew into stages 2 and 4; Chapter 25's mono doctrine and Chapter 28's M/S became the width section's law and the side high-pass; Chapter 26's drive-and-match returned as crumbs; Chapter 28's gain-reduction stacking became limiter staging; Chapter 30's protocol became translation insurance; and Chapter 31 defined the job this chapter executed — polish, not surgery, with the print spec that made it possible. Even Chapter 2 finally paid its strangest debt: the not-stairsteps threshold concept is why true peaks exist between samples.

Forward: Chapter 33 puts numbers on what you just did by ear — LUFS, normalization, and why the slammed master is a casualty of a war that already ended. Chapter 35 takes your printed WAV and its metadata to the platforms. And Chapter 38 runs the experiment Jaylen is already curious about: his Chapter 32 master versus an AI mastering service, blind.

Spaced Review

From Chapter 31: A bandmate says "mastering will fix the buried vocal, right?" — what's the honest answer, and what's the print spec you'd hand a mastering engineer (or your own master chain)?

Verify

No — mastering is quality control and translation, not mix surgery; a buried vocal is a balance problem and goes back to the mix (the Chapter 20 hierarchy). The print spec: 24-bit WAV, peaks around -6 dBFS, no limiter on the 2-bus, fades decided or left for mastering, alts labeled — delivered after the next-day listen and one focused revision pass.

From Chapter 30: Name the five translation-test systems and one thing each reveals.

Verify

The car (low-end truth and real-world level in context), the phone speaker (does the low end imply via harmonics; does the lead survive), earbuds (sibilance and harshness up close), the quiet listen (does the vocal still lead; do the important elements survive at whisper level), and the next room (overall balance truth — what pokes out when you can't see the meters).

From Chapter 28: State the gain-reduction stacking doctrine and why it works.

Verify

Several gentle stages beat one violent one — e.g., 2 dB from each of three stages over 6 dB from one — because each stage operates inside its own transparent zone, below the threshold where its artifacts become audible, and small artifacts from separate stages don't compound the way one overworked circuit's panic does. This chapter applied it to the final gate as limiter staging.

What's Next

You just set your record's loudness by ear, against references, at matched loudness — which means you're ready for the chapter that explains the war fought over exactly that knob. Chapter 33 is the peace treaty: the loudness war's history and casualties, LUFS and true peak as proper measurements, and the genuinely liberating fact that streaming normalization turns everything down to match — so the slammed master arrives with its punch already spent, and dynamics are a choice again. Jaylen's about to discover what Spotify does to "the loud one." Do this chapter's exercises first — especially the 0.5 dB ladder — because the next chapter lands hardest on ears that have just felt the edge themselves.