Chapter 31 Exercises

This chapter's lab is unusual in two ways. First, almost nothing in it requires a tool you don't already own — the work is decisions, rituals, and writing, which is exactly what the pre-mastering stage is made of. Second, the centerpiece can't be rushed: the next-day listen is literally next-day, so exercises C1 through C3 span at least three calendar days by design. Read the whole set before you start anything, then put the listens on a calendar like the appointments they are.

You'll need: your diagnosed, finished mix from the Chapter 30 checkpoint (if the symptom table hasn't been run, run it first — printing an undiagnosed mix is gift-wrapping problems), your three reference tracks from Chapter 4, your full translation fleet (main monitoring, car, earbuds, phone), your listening journal, and a calendar. One law governs every comparison in this set, and it matters more here than anywhere else in the book: matched loudness, or the comparison is void. The entire mythology of mastering runs on unmatched A/Bs. Don't run one on yourself.

Exercises marked ★ have worked answers in the back of the book (Answers to Selected Exercises). 🎧 marks focused listening; 🎛️ marks exercises that need your DAW open.

Part A: Conceptual Understanding

A1. The six jobs, cold. ★ (10 min) From memory: name mastering's six jobs and write one sentence each on the question it answers. Then the two follow-ups the chapter built its table for: which single job was Theo performing at 1 a.m. — and what was missing in front of it? And which job is structurally impossible to perform on your own record — not for lack of skill, but for lack of what?

A2. The colorist and the proofreader. (15 min) The chapter gave you two analogies from other media. Map them: which of the six jobs does the film colorist's work explain, and which does the proofreader's? Then find the seams — name one real mastering job that neither analogy covers well, and one way each analogy could mislead someone who leaned on it too hard. (Hint for the colorist: what does a colorist get to do shot by shot, scene by scene, that an engineer holding a single stereo file cannot?)

A3. Polish, not surgery — the mechanism. (15 min) Write the four "mastering cannot" statements from memory, and for each, explain why in terms of the one structural fact that generates all of them: mastering works on the sum. Use the vocal-and-cymbals example — or build your own pair of instruments sharing a frequency zone — to show why "bring the vocal up" is not an instruction a mastering EQ can actually execute.

A4. The detector, calibrated. ★ (20 min) Classify each note as a mix note (names an element — travels upstream to Chapter 30's method) or a mastering note (names the whole), with one clause of reasoning per call: (a) "The hook vocal is buried in the double chorus." (b) "The whole record reads dull on top next to the references." (c) "Track 3 comes in noticeably louder than track 2." (d) "The snare's ring pokes out in the bridge." (e) "The gap before the closer feels rushed." (f) "Earbuds make the top end spit." (g) "The 808's first note clips somewhere." (h) "The low-mids feel thick through every chorus." Two of these are genuinely borderline — say which two, and what extra information would settle each. The detector routes 95% of notes correctly; the reasoning is for the other 5%.

A5. Autopsy of the magic-sauce myth. (15 min) The threshold block named three reasons the mastering-as-magic myth survives every debunking. List them. Then explain what a matched-loudness A/B does to the first one — and why the third reason (great masters descend from great mixes) is, on reflection, the most encouraging sentence in the chapter for someone who just finished Parts V and VI.

A6. The spec card, line by line. (20 min) Choose five lines of the mix-print spec card and, for each, describe the specific failure it prevents — the thing that goes silently wrong and surfaces later on someone else's speakers. At least one of your five must be the head/tail line (what exactly is lost when a reverb tail gets truncated, and why is it unfixable downstream?) and one must be the 2-bus rule, both halves: what comes off, what stays, and why stripping the Chapter 26 color delivers "a mix nobody approved."

A7. Stem mastering, named honestly. (10 min) Define stem mastering in one sentence, list what it buys and what it costs (T3 — nothing is free), and explain the chapter's verdict that it's "a remix with a nicer suit on." Then describe the one situation where you'd genuinely consider it, and what you'd want to be honest with yourself about before paying for it.

A8. The four postures. (15 min) The mastering mindset section named four shifts: whole-song moves, the quarter-dB world, consume-first, and system-and-room trust. For each, write one sentence on why the mixing posture — zoom in, isolate, interrogate, fix — actively fails at the mastering stage. Then connect the fourth shift to the honest answer to "what am I actually paying a mastering engineer for?"

Part B: 🎧 Listening Lab

Every comparison below happens at matched loudness unless the exercise explicitly says otherwise — and where it says otherwise, the mismatch is the lesson. Journal open, timestamps on everything, and split this part across two sittings; ears judging quarter-dB differences fatigue faster than ears judging EQ moves.

B1. The Demi drop, reproduced. ★ (30 min) Build the playlist Demi built in the chapter's opening: your freshly printed mix (exercise C1's output, or any finished print you have) shuffled among your three Chapter 4 references. Round one, deliberately unmatched: one volume setting, main system, let the shuffle cross the borders. Journal the drop the way Demi would — not just "quieter," but quieter and what else? Density? Top-end sheen? Tonal kinship with its neighbors? Does yours sound like it's asking permission? Round two, matched: trim your print's playback gain until it sits level with the references, and listen again. Whatever difference survives the match is real — that's your actual gap, and most of it belongs to Chapter 32. Write the two lists side by side: what the level difference was pretending was wrong, and what's genuinely different.

B2. Dismantle a demo. (25 min) Find any mastering before/after demo online — a plugin product page, an online mastering service's player, a YouTube "watch me master this" video. Listen exactly as presented and write your first impression in one line. Now level-match the pair (import both into your DAW and trim, or ride the volume carefully) and A/B again. Journal what the "magic" shrank to, named in band vocabulary: a half-dB of air? Some glue? A touch of width? You're not hunting fraud — most demos are sincere. You're calibrating how much of every unmatched "after" is Chapter 4's louder-sounds-better, working exactly as advertised.

B3. Album kinship audit. (30 min) Pick two records that matter to you — one in your genre, and one anyone's; When We All Fall Asleep, Where Do We Go? works beautifully here. On each, listen to three consecutive track seams (the last thirty seconds of one song into the first thirty of the next). Log per seam: did your hand want the volume knob? Does the tonal family hold — does the later track read brighter or darker than its neighbor for no musical reason? How long is the gap (count it out), and does its length match the emotional pivot? You're auditing jobs 1 and 4 on finished records: hearing the invisible work by confirming you can't hear it.

B4. Sequencing on named records. ★ (40 min) Three stops. Abbey Road, side two: from "Here Comes the Sun" into the medley — map where songs touch with no air at all, where a breath of silence survives, and where one track simply becomes the next. The medley is the most famous sequencing-and-spacing argument ever pressed to vinyl. The Dark Side of the Moon: pick any two adjacent tracks and try to mark the exact moment one ends; the record is built so you mostly can't, which is job 4 promoted into a compositional device. The quiet closer: "goodbye" ending When We All Fall Asleep, Where Do We Go? — or "Something in the Way" closing Nevermind — sits audibly below its album's loud songs. Journal: what would be destroyed if mastering had limited the closer up to match? Connect your answer to the chapter's line that level relationships between songs are musical information.

B5. The quarter-dB drill. ★ (30 min, repeatable) The chapter promised this one: training your ears toward mastering resolution. Prep (DAW): duplicate one reference track and put a +0.5 dB high shelf around 8 kHz on the copy; re-check overall loudness (a small shelf barely moves it, but verify); bounce both with neutral names. Test (ears only): blind A/B — shuffle the files, or have a friend drive — ten trials, hit rate logged honestly. Above eight of ten? Rebuild the copy at +0.25 dB and repeat. Below six? Go to +1 dB, find your current floor, and write it down without shame — a floor you know is worth more than a sensitivity you're guessing at. Then run one round with a +1 dB low shelf at 100 Hz and notice how differently low-end increments read on your monitoring (Chapter 10 has opinions about why). Re-run monthly. The floor drops with practice, and that floor is your readiness for Chapter 32's two-dB world.

B6. Gap forensics. (20 min) Two albums, very different moods. For four consecutive seams on each, time the silence between tracks — count it, stopwatch it, or import into your DAW and read the gap off the timeline. Log each number next to what's happening musically at the seam. Where do gaps stretch, and is it reliably before the quiet, heavy, or strange song? Where do they vanish to nothing? Close with one sentence on what the gap before a song does to that song's first ten seconds.

B7. Translation, achieved. (25 min) Run one commercial master in your genre through the full Chapter 30 lap: main system, car, earbuds, phone speaker, quiet listen at conversation-breaking volume. Log what each system changes (tone, width, how much sub survives) and what stays untouchable (the balances, the vocal's seat, the depth order). That untouchable layer is what "translates everywhere" sounds like after professional QC — and it's the baseline your own print's lap in exercise C2 gets judged against. One closing line: which of your systems told you the least, and do you trust it less now?

Part C: 🎛️ DAW Lab

C1. Print to spec. ★ (60 min + one verification listen) The chapter's core deliverable — the Fast Track sends people straight here for a reason. Execute the spec card on your finished mix, in order. (1) 2-bus audit: anything whose job is loudness comes off (the reference-check limiter, any maximizer or clipper earning level); sound-shaping glue and color from Chapter 26 stays — the mix was balanced through it. Genuinely unsure about a processor? Print both versions, label both honestly. (2) Format: 24-bit WAV, session-native sample rate, no conversion at print. (3) Level: confirm peaks land around -6 dBFS (anywhere in the -3 to -10 range is honest); if you're staring at peaks near zero instead, stop — that's a Chapter 21 trim-gain audit, about twenty minutes, balances untouched. (4) Head and tail: roughly one second of clean silence up top; every reverb and delay tail rings fully out. (5) Alts, same pass, same settings: instrumental, a cappella, vocal-up +1 dB, clean version if relevant. (6) Names: Artist_Song_MixPrint_v1_24-48.wav and family — "final" remains a banned word. (7) Verify: one full top-to-tail listen of the actual rendered file. The render is not the session. Deliverable: the print folder, plus one journal line per spec row confirming compliance — the card becomes reflex by being checked, not admired.

C2. The next-day listen, run as written. (two days; about 60 min of listening) Tonight: nothing. That's the exercise — do not evaluate the print on print night. Tomorrow, or better the day after: Pass one — consume. Main system, calibrated level, DAW closed (non-negotiable; an open session turns you back into a mixer). The whole record, no skipping, attention on the experience. Pass two — translate. Car, earbuds, phone, quiet listen. Use this template in your journal, verbatim or adapted:

NEXT-DAY LISTEN — [song] — printed [date], heard [date]
SYSTEM / LEVEL: [main system, calibrated level]

PASS 1 — CONSUME (DAW closed):
  Experience verdict (one line — did it move, where did attention drift):
  [m:ss] — [symptom, band vocabulary] — [element or whole?]

PASS 2 — TRANSLATE:
  Car:     [m:ss] — [symptom] — [element/whole]
  Earbuds: [m:ss] — [symptom] — [element/whole]
  Phone:   [m:ss] — [symptom] — [element/whole]
  Quiet:   [m:ss] — [symptom] — [element/whole]

TRIAGE:
  PROBLEMS    (objective, nameable, would embarrass you later):
  PREFERENCES (different-not-better — written down, let go):
  MASTERING NOTES (whole-record — these go in the brief):

ONE FOCUSED PASS: scheduled [date] — list locked.

Deliverable: the completed template. Then grade your own notes against the chapter's standard — timestamped, band vocabulary, symptom-first. "Something's off in the middle" earns itself a rewrite.

C3. The revision notes, then the one focused pass. (30 min writing + one session) First, the writing drill: pull the three vaguest notes from your C2 template — every first template has them — and rewrite each to engineer grade: timestamp, symptom in Chapter 4's band vocabulary, element-or-whole verdict, and the stage the fix belongs to. ("Chorus feels harsh" becomes "2:10–2:31 — hat/vocal sibilance stack, 5–8 kHz, earbuds only — element — mix: de-ess or hat shelf.") Then triage the full list into the three buckets and execute: one session, the problem list and nothing but the problem list, every fix verified at matched loudness, no wandering. Re-print v2 per the full spec card; run one verification listen of the new file. Close the journal entry with the stop-condition, honestly: either "the remaining notes are all preferences — done," or, if the problem list ran long and kept naming elements, the truer sentence — "this wasn't a revision; the mix wasn't finished" — and a scheduled date with Chapter 30's method. Both endings are wins. Only one of them is comfortable.

C4. Theo's experiment, failed safely. ★ (30 min) The chapter promised you'd get to do this once, on purpose, where it can't hurt anything. In a fresh session — never your mix session — import a copy of your C1 print. Strap your stock limiter across it and do exactly what Theo did: drag the threshold down until the meter shows about 8 dB of gain reduction on the peaks. Bounce it. Now the step Theo skipped at 1 a.m.: turn the loud version DOWN until it matches the original print, and A/B at identical loudness. Log the casualties in band vocabulary: what happened to the transients — snare crack, hat ticks, pick attack? The air you protected through Part VI? The depth — does the back of the stage still exist? The quiet sections — can you hear the mix swell and duck, the pumping Theo heard in the waltz's bridge? Then find the onset: raise the threshold in 1 dB steps and log the gain-reduction figure where the damage stops being audible to you, on your monitoring. Write that number down — it's a preview of the negotiation Chapter 32 conducts with better tools and a gentler hand. Delete the session when you're done. The print survives untouched, and the limiter now holds no mystery and no temptation.

C5. The hire-or-self worksheet. (40 min, no DAW) Walk the chapter's decision tree on paper for your actual next release — not a hypothetical. Write answers, not vibes. Stakes: what is this release carrying? Statement EP, or single eleven in a steady cadence? Anything pressed to vinyl or CD, where errors become permanent objects in boxes? Monitoring truth: what's the lowest frequency your room reports honestly — write the number you actually believe after your Chapter 8 and Chapter 10 work — and does your genre live below it? Distance: how many listens deep are you, and how many weeks can the release wait? Budget: the real number you could spend without resentment. Then research, without contacting anyone yet, two or three working mastering engineers in your genre lane: published rates, turnaround, what their delivery spec asks for, whether revisions are included. Render the verdict the tree renders. Then write the hybrid-play plan regardless of the verdict: which future release gets the professional treatment, and what you'll do with the A/B when it happens. Deliverable: a one-page mastering policy for your next twelve months, filed where you'll find it.

C6. The brief, written for a stranger. (30 min) Write the full mastering brief for your track — even if, especially if, the engineer receiving it is you in Chapter 32. Contents: your three references, each with one sentence naming what it demonstrates ("the low-end weight of A, the vocal intimacy of B"); your intent in feel-words tied to moments, not adjectives floating free; your confessed concerns, honestly stated ("our room lies below 60 Hz — the bridge's low end is a guess"); the tech doc (sample rate, bit depth, peak level, tempo, key, runtime); your fade and tail intentions; the alt list. File it in the print folder. The test of a good brief: a competent stranger could master your record closer to your intentions with it than without it. Chapter-32-you is exactly that stranger.

Part D: Synthesis & Critical Thinking

D1. The cheapest consultation in audio. (20 min) Defend the chapter's claim that a mastering engineer sending your mix back with notes is the highest-value mix feedback you'll ever receive. Build the argument from three sources: Chapter 19's feedback rules (what makes notes actionable), job 6 (what fresh, trained, full-range ears catch), and the four hire factors. Then steelman the other side: name two real reasons a working producer on a steady release cadence might rationally never hire out — and reconcile the two positions in one closing sentence.

D2. Mastering after the war. (25 min) Chapter 33 tells the full story, but reason ahead from this chapter's table: if every major platform normalizes playback loudness — turning the slammed master and the dynamic master down to the same level — which of the six jobs lose value, which hold steady, and which arguably gain? Write a paragraph predicting what the profession looks like once loudness is a solved deliverable instead of an arms race. Keep your reasoning where you can find it; check it against Chapter 33 when you arrive.

D3. A perception question, not a money question. (20 min) The chapter's Deep Dive routing called the self-master-or-hire decision "a perception question instead of a money question." Argue both sides. First, make the sidebar's case: the room's bandwidth, the fixed reference level, years of full-attention comparative listening — things no plugin purchase replicates. Then make the budget-reality case the chapter equally insists on: the gap is knowledge, not money, and every self-master is a rep that trains your ears. Finish by naming the release scenarios where money genuinely is the deciding variable — and the ones where it genuinely isn't.

D4. Engineer the proofreader. (25 min) Job 6 — the last set of ears — cannot be self-performed: you hear your memory of the file, not the file. Design the closest substitute protocol you can actually run: how many days of rest, which system rotation, what hunt-checklist (clicks, edit seams, fades that end abruptly, channel swaps, truncated tails), and one borrowed pair of ears with a Chapter 19-structured ask. Write it as a numbered procedure you could hand to your future self on a release week. Then, in one honest closing paragraph, name what your protocol still cannot replace — and why that residue is the strongest single argument living on the "hire" branch of the tree.

Part M: Mixed Practice

M1. (Chapter 2) The dither doctrine. ★ (10 min) Your print is 24-bit with no dither; your eventual CD master will be 16-bit with dither. Explain the ordering: why is dither unnecessary at the 24-bit print, when does it finally earn its keep, and why does the chapter insist that rate and bit-depth conversions happen once, at the deliverables stage, instead of casually along the way? Bonus: a friend wants to send their mastering engineer a 320 kbps MP3 "because it sounds identical." Using Chapter 2's vocabulary, explain what's permanently missing and why mastering it means polishing the artifacts.

M2. (Chapter 4) The oldest trick, in new clothes. (10 min) An online mastering service's "after" file plays about 3 dB louder than your upload and "obviously sounds better." Name the Chapter 4 mechanism being exploited, design the honest comparison in two steps, and predict — in band vocabulary — which "improvements" will survive the match and which will evaporate.

M3. (Chapter 21) The hot print. (15 min) You go to print and find 2-bus peaks at -0.4 dBFS — no clipping yet, but no headroom either. A bandmate says "pull the master fader down 6 dB and bounce." Explain why that's only half a fix when there's processing on the 2-bus (what level are those processors actually being driven at, and does the master fader change it?), describe the Chapter 21 trim-gain audit that fixes it properly, and state what must remain untouched while you do it. Levels are relationships — and the relationships are finished.

M4. (Chapter 19) Notes triage. (15 min) Rewrite each hurting note into a helping one, or route it where it belongs: (a) "Make it slap more." (b) "Use a multiband around 120 Hz on track 2." (c) "Bring my guitar up a touch on track 3." For each: name what's wrong with the original — no anchor? prescribes the engineer's tools? wrong domain entirely? — then write the version a professional could act on, or the sentence that sends it upstream instead.

M5. (Chapters 23 + 26) The 2-bus eviction hearing. (10 min) Five processors sit on a 2-bus at print time. Rule on each — stays or comes off — and cite the principle: (a) subtle tape emulation the whole mix has been balanced through since Chapter 26; (b) a glue compressor doing 1–2 dB of slow gain reduction, also there for weeks; (c) a maximizer added last night "to check against references"; (d) a brickwall limiter at -0.1 dB "for safety"; (e) a mid-side EQ making a 0.5 dB side cut that solved a width problem back in Chapter 28. One of the five is genuinely arguable — say which, and what you'd do about the argument. The spec card has a rule for exactly this situation.

Selected answers and worked examples: see the Answers appendix. The spine of this set is C1 → C2 → C3 — print, rest, revise — run on real calendar days; if the week only gives you those three, take them in order and call it a win. And keep the chapter's closing promise: C4 lets you fail Theo's experiment once, on a copy, on purpose — and never again on a deadline.