Chapter 28 Quiz

Closed book on the first pass — that's the honest score. Multiple choice, then true/false with justification (the justification is most of the point), then short answer, then one applied scenario: an over-processed mix on the table and you holding the delete key. Answers hide under each Verify fold; the scoring guide is at the end.

Section 1 — Multiple Choice (2 points each)

1. Sidechain compression, in one sentence:

A) Compression applied to a copy of the signal instead of the original B) Ordinary compression where the detector listens to a different signal than the one the gain element turns down C) A compressor with two thresholds D) Compression that only affects one frequency band

Verify **B.** That's the entire trick. The compressor is still the [Chapter 23](../../part-05-mixing-foundations/chapter-23-compression/index.md) two-part machine — detector plus gain element — but the detector's input (the key) is patched to another signal. Sound A decides when sound B yields.

2. In the kick→bass duck, what happens to the kick's audio?

A) A little of it blends into the bass channel through the key input B) It's rerouted through the bass compressor and back out C) Nothing — the key line is a copy feeding the detector only; the kick's audio path is untouched and no kick audio enters the bass channel D) It's phase-inverted to avoid masking

Verify **C.** The kick is *information* on that channel, not sound. First-timers expect to hear kick bleeding through the bass track; what you hear instead is nothing at all until both sounds play together — which is the point.

3. The transparent kick/bass duck recipe at roughly 96 BPM:

A) 4:1, fastest attack, release ~90 ms, threshold for 2–4 dB per hit B) 10:1, slow attack, release ~400 ms, 10 dB per hit C) 2:1, 30 ms attack, release ~625 ms, 6 dB per hit D) Any settings, as long as the meter never moves

Verify **A.** Fast attack so the bass is out of the way at the instant of impact; release short enough that the bass is home before the next subdivision that matters (an eighth note at 96 BPM is 312 ms, so ~90 ms has room to spare); 2–4 dB so you hear the *kick improve*, not the bass move.

4. You set the duck's release to 400 ms at 96 BPM without meaning to. What you hear:

A) Nothing — release doesn't affect transparency B) The bass is still climbing back when the next subdivision arrives, so the recovery itself becomes audible — an unchosen pump, seasick instead of groovy C) Distortion on the kick D) The kick gets quieter

Verify **B.** The recovery motion is the difference between the two aesthetics. Short release: home before anyone notices, transparent. Long release: the climb back is audible — which is an *aesthetic* when chosen and timed to the groove, and a lurch when it isn't.

5. The ghost trigger is:

A) A reverb tail that triggers the compressor B) Keying the duck from a muted track playing a steady four-on-the-floor pattern, so the pump keeps breathing through breakdowns, survives kick-sample swaps, and stays even under a syncopated audible kick C) A second compressor in series D) An automation lane that fakes the pump

Verify **B.** The detector just needs a clock — the key signal and the musical kick don't have to be the same thing. Once you control what the detector hears, you control what the compressor believes.

6. The reverb-return duck uses a slower attack (10–30 ms) than the kick/bass duck because:

A) Reverbs can't be compressed quickly B) It's protecting intelligibility across phrases, not a transient — the relaxed attack lets the verb bloom onto each phrase's onset before yielding, so the space feels continuous instead of gated C) Slow attacks use less CPU D) The vocal needs time to trigger the detector

Verify **B.** The kick/bass duck protects an impact, so it must be instant. The space duck protects words. Different job, different time constants — same two-part machine.

7. Parallel (NY) compression is:

A) Two compressors in series with opposite settings B) Sending a copy of the signal to a compressor set to crush, then blending the crushed copy underneath the untouched original C) Compressing only the side channel D) Compressing the left and right channels separately

Verify **B.** Not instead of — *underneath*. The dry path keeps every transient at full size; the wet path supplies the dense, sustaining floor; the return fader is a density knob.

8. Parallel compression preserves punch while adding density because:

A) The compressor is set to a low ratio B) At loud moments the sum is dominated by the untouched dry path, and at quiet moments by the crushed path — peaks pass essentially unchanged while the valleys get lifted: upward compression by routing C) The crushed copy is delayed slightly D) Parallel compressors use a special detector

Verify **B.** Adding a much quieter copy under a loud peak barely changes the peak; meanwhile the crushed copy's quiet moments sit nearly as loud as its loud ones, so the valleys of the blend rise. Your ear judges punch by peaks (untouched) and richness by what's audible between them (newly lifted).

9. You build a parallel drum lane and the summed drums get thinner and hollow. Most likely cause and the sixty-second diagnostic:

A) Too much gain reduction; lower the ratio B) The two paths are out of time and comb-filtering — polarity-flip the return at equal level: a deep null means aligned, a thin phasey wash means a latency problem to fix before blending C) The send is post-fader; switch it to pre-fader D) The crush compressor's knee is too soft

Verify **B.** Same audio on two routes plus a latency difference equals a comb filter. Modern plugin delay compensation handles it almost always — which is why the polarity-flip audit matters for the times it doesn't.

10. Mid-side decomposition:

A) M = left channel, S = right channel B) M = L + R (everything the channels agree on — the mono sum); S = L − R (everything they disagree on — the differences that create width); mono playback hears M alone C) M = the midrange frequencies, S = the sides of the room D) A widening effect that adds artificial reflections

Verify **B.** Same audio, different coordinate system — and the decode (L = M + S, R = M − S) rebuilds the stereo file losslessly. The mono fold is the M channel by definition, which is why side-channel level moves are loans mono never repays.

11. The single most useful M/S EQ move in mixing, per the chapter:

A) Boosting the mid channel's low end for power B) High-passing the side channel (somewhere around 120–250 Hz) — width-flavored mud leaves, the centered low end is untouched, the lows go effectively mono, and the image tightens instead of narrowing C) Cutting the mid channel at 1 kHz D) Boosting the sides broadly by 3 dB for width

Verify **B.** The sides of a dense mix usually hide a swamp — pad low-mids, stereo reverb bottom octaves, detune spread — that does nothing for width and everything for blur. Vacuuming it out is the workhorse; D is the seductive mistake.

12. A dynamic EQ band differs from a static cut because:

A) It has a steeper slope B) It sits flat — asleep — until energy at its frequency crosses a threshold, dips by up to its allowed range, then returns: you pay the cut's cost only during the moments the problem actually exists C) It only works on vocals D) It cuts harder at higher frequencies

Verify **B.** A parametric band with a tiny compressor inside. The decision ladder: constant problem → static EQ; intermittent problem within one signal → dynamic EQ; two signals colliding in time → sidechain; per-band dynamics across a whole mix → multiband, with supervision, in Chapter 32.

13. Reaching for a multiband compressor in a mix usually means:

A) You've outgrown single-band compression B) Something upstream is unsolved — a bass that needs multiband leveling needs Chapter 23 or Chapter 15; fighting bands need Chapter 22 or Chapter 16. It's a mastering tool that occasionally moonlights, and legitimate use comes with a one-sentence justification C) Your converters are low quality D) The mix is ready for release

Verify **B.** Four-plus compressors in a trench coat, plus crossover points that smear phase right where you put them. Exhaust the ladder first; when the moonlighting is legitimate (a printed stem you can't re-balance), you'll be able to say why in one sentence.

14. Mute each of these mid-song. Which species behaves how?

A) Group: drums survive, smaller. Parallel: drums gone. Return: room gets wetter B) Group: the drums are gone (tracks flow through it). Parallel: the drums survive, smaller (it runs beside them). Return: everything keeps playing, in a drier room (it's wet-only weather on top) C) All three behave identically D) Group: drier room. Parallel: drums gone. Return: drums smaller

Verify **B.** Groups carry, parallels season, returns weather. When a session confuses the species, every later decision gets harder — and the stem prints inherit the confusion.

15. Your 2-bus compressor pumps audibly on every kick even at 1–2 dB of gain reduction. The fix that doesn't touch the mix's actual low end:

A) High-pass the whole mix at 100 Hz B) High-pass the detector — the sidechain filter — around 100–150 Hz, so the compressor stops counting bass energy when deciding to act while the audio path keeps its lows C) Slow the attack to maximum D) Switch to a multiband compressor

Verify **B.** The quiet workhorse of the detector-circuit sidebar. The audio keeps its lows; only the *measurement* ignores them. It's the difference between a bus compressor that breathes with the song and one that flinches at every kick.

Section 2 — True/False with Justification (2 points each: 1 for the call, 1 for the why)

16. The key input mixes a small amount of the kick's audio into the bass channel — that's how the duck knows when to act.

Verify **False.** The key feeds the *detector*, a measurement circuit that hears everything and touches nothing. No kick audio enters the bass path at all; the kick is information there, not sound. The audio you hear passes only through the gain element.

17. A snare running through a 4 dB punch compressor and 2 dB of bus glue, with a parallel crush running 10 dB beside it, is subject to roughly 16 dB of serial gain reduction.

Verify **False.** About 6 dB serial. The crush's 10 dB happens on a parallel copy summed *beside* the dry path — it never processes the signal the listener's snare transient travels through. Parallel gain reduction doesn't stack in series; that's why the lane is the gentle option despite its brutal meter.

18. Boosting the side channel's level is a free width win, since mono listeners only hear the M channel anyway.

Verify **False — and the "anyway" is the trap.** Every decibel added to S exists for stereo listeners and vanishes for mono ones, so the two experiences drift apart: the mix collapses *more* in mono relative to its stereo self, and the correlation meter slides toward zero. Narrow, chosen side moves (the air shelf) are legal with a mono check after every one; broad S boosts are width bought with energy half your audience never receives.

19. In a mix context, wanting a multiband compressor is usually a symptom that an upstream problem went unsolved.

Verify **True.** A bass needing multiband leveling needs [Chapter 23](../../part-05-mixing-foundations/chapter-23-compression/index.md)'s compressor or [Chapter 15](../../part-03-recording/chapter-15-editing/index.md)'s editing; fighting frequency zones need [Chapter 22](../../part-05-mixing-foundations/chapter-22-eq/index.md)'s EQ or [Chapter 16](../../part-04-arrangement-production/chapter-16-arrangement/index.md)'s arrangement. The tool stacks decisions faster than almost anything in the rack. Its honest home is mastering ([Chapter 32](../../part-07-mastering/chapter-32-mastering-tools-techniques/index.md)), where you can't reach into the mix and fix the source.

20. When a candidate fails the bypass test, the prescribed move is to leave it bypassed in the session in case you want it later.

Verify **False.** Delete it at the end of the pass — bypassed plugins are clutter with a pilot light, and they keep charging complexity tax (diagnostic fog, session rot, stem-print confusion) without paying rent. The one exception: write the *idea* into the MIX NOTES file with a date. Ideas are free. Inserts are not.

Section 3 — Short Answer (4 points each)

21. Write out the bypass-test ritual's five steps in order, including the exact question step three asks and why that wording matters.

Verify (1) Full mix, busiest section, matched loudness — solo banned, level bias handled. (2) Listen with the candidate in for fifteen to thirty seconds, then bypassed, the same stretch — twice each minimum, because first impressions favor whatever changed. (3) Ask: **is the mix *worse* without it?** — not "can I hear it?", because you can hear anything once you know it's there; "worse" forces the technique to defend the song, not your effort. (4) Say what it's doing, out loud, in one sentence; "it adds vibe" is the tell. (5) No sentence, no insert: delete it (don't park it bypassed), with the MIX NOTES exception for ideas worth a dated line.

22. Name the three species of bus. For each: the question it answers, what its fader means, and one example from the "Static Bloom" architecture.

Verify **Group (family) bus** — *where does this audio live?* Tracks flow through it; its fader is the level of the whole family. Example: DRUMS (or BASS, SYNTHS, GTR, VOX, FX). **Parallel bus** — *what runs alongside it?* A copy runs beside the original and sums back in; its fader is the dose of the treatment. Example: DRUM CRUSH. **FX return** — *what world does it sit in?* Wet-only signal added on top, fed by sends; its fader is the amount of shared space. Example: PLATE (or ROOM, TAIL, the delays). Mute test: group muted = family gone; parallel muted = family survives, smaller; return muted = everything plays, drier.

23. State the gain-reduction stacking doctrine in one line, give the three compounding reasons it wins, then name its failure mode and the audit that catches it.

Verify Doctrine: **2 dB three times beats 6 dB once** — distribute total dynamic control across several gentle serial stages. Why: each stage works in its comfort zone (at 2 dB the envelope barely bends and the tonal thumbprint stays faint); each stage gets one job with the right time constants (track levels, bus glues, master breathes); and errors don't compound the same way (a slightly-wrong 2 dB stage is a rounding error the next stage absorbs; a slightly-wrong 6 dB stage is the sound of the record). Failure mode: *stacking you didn't choose* — five uncounted 2 dB stages is 10 dB of squash with no single culprit. Audit: play the chorus, walk the signal path on the map, watch every GR meter in series, add them up out loud; if the sum surprises you, find the freeloading stage with the bypass test.

24. "A compressor, a ducker, a de-esser, a dynamic EQ, and the four-on-the-floor ghost pump are all the same machine." Defend that sentence using the detector concept, and state the one question that differentiates them.

Verify Every dynamics processor is two parts: a detector that measures level and issues a control instruction, and a gain element that does the turning down — the audio passes only through the gain element. The differentiating question: **what does the detector listen to (and what does the gain element act on)?** Same signal it controls → compressor. A different signal → ducker/sidechain. A filtered copy of itself → frequency-conscious compressor (sidechain HPF). A narrow band, acting on that band → dynamic EQ; parked on sibilance → de-esser ([Chapter 29](../chapter-29-mixing-vocals/index.md)). A muted ghost track → the pump that breathes through breakdowns. The uniforms change; the machine doesn't.

Section 4 — Applied Scenario (10 points)

25. A friend sends you their "advanced" mix for help, and the session is a museum of this chapter applied without the conscience: (a) every track is sidechained to the kick because a tutorial said so — the whole mix inhales on every hit, including the lead vocal; (b) the parallel drum lane makes the drums thinner than bypassing it; (c) the crush keeps blasting at full level through the quiet outro even though the drum fader rides down; (d) the mix is wide and impressive on their monitors and collapses to mush on every phone; (e) the lead vocal carries six dynamic EQ bands, three of which are visibly working on every single phrase; (f) nobody — including your friend — can predict what bypassing any one plugin will change. Diagnose all six: name each problem and its fix, then state the order you'd work in and which ritual question kills which insert.

Verify Order: **draw the map first** (f is the meta-problem — you can't audit routing you can't see), then fix the mechanical faults, then run the ritual at matched loudness on everything that remains. **(f) Architecture spaghetti.** Draw the full map, one page, three species, dashed keys. Lanes that can't be explained in a sentence get merged or deleted. This converts the session from archaeology site to diagram — and makes every following step possible. **(b) Comb filtering.** The parallel paths are out of time. Polarity-flip the return at equal level: no deep null = latency problem; repair plugin delay compensation (or remove the offending plugin) before any blend judgment. **(c) Pre-fader send.** The crush send ignores the drum fader. Switch it post-fader so the lane follows every fader and automation move, then re-set the blend from silence in the full mix. **(a) Tutorial-itis ducking.** Remove the ducks pairwise, keeping only collisions you can *name* — kick/bass, vocal/space. A duck on the lead vocal keyed from the kick fails the sentence test on contact (what relationship is it managing?). Expect to delete most of them. **(d) Width bought on credit.** Hunt the broad side-channel boosts and the parallel wideners; undo them. Keep only narrow, chosen side moves (the air shelf), re-run the correlation meter and the mono check after each, and accept that the stereo image may get slightly smaller and the *record* better. **(e) Dynamic EQ as fancier static EQ.** Bands that work on every phrase aren't intermittent — convert those three to [Chapter 22](../../part-05-mixing-foundations/chapter-22-eq/index.md) static cuts (cheaper, more predictable); keep at most the genuinely conditional band(s). The ladder: take the lowest rung that solves the problem. Then the full ritual on every survivor: full mix, busiest section, matched loudness, in/out twice, the one-sentence test — "is the mix worse without it?" deletes the freeloaders. Full credit needs: the map first with a reason, the phase and pre/post-fader mechanics correctly named, the pairwise duck removal with the named-collision rule, the mono/correlation discipline on width, the dynamic-to-static conversion logic, and the ritual's question quoted accurately.

Scoring

Section Points available
Multiple choice (1–15) 30
True/False + justification (16–20) 10
Short answer (21–24) 16
Applied scenario (25) 10
Total 66

56–66: You know what the detector hears, what the fader means on all three species, and which question kills a freeloading insert. Lock your architecture and go meet the vocal chain in Chapter 29. 44–55: The tools are in; the conscience is wobbly — re-read "The Architecture Core" from the three-species table through the ritual, then redo exercises A6–A8 and C7. Below 44: No shame; this chapter compresses four techniques and a philosophy into one number. Rebuild the duck (C1) and the crush (C3) with the chapter open — the routing teaches the concepts better than the prose does — then retake this in two days.