Chapter 9 Exercises

Work these roughly in order: Part A locks the concepts, Part B trains your ear to hear programming on records, and Part C is where the chapter becomes muscle memory — the DAW Lab assumes you've done at least B1 and B2. Time estimates are honest averages. Selected exercises (marked ★) have worked answers in the back of the book (Answers to Selected Exercises). One ground rule for everything below: judge results at your Chapter 8 reference level, through monitoring whose lies you've documented. Velocity-balance decisions made at random volumes are random decisions.

Part A: Conceptual Understanding

A1. The script and the actor. ★ (10 min) A friend emails you a .mid file of their new song and asks "how does the piano sound?" Explain, in four sentences or fewer, why the question doesn't parse — covering what the file actually contains, where sound enters the signal flow, and what they'd need to send instead for you to hear their version. Then name one practical superpower and one practical hazard of this instructions-not-audio architecture.

A2. Velocity's two jobs. ★ (10 min) Explain why velocity on a well-built virtual instrument is "a timbre selector wearing a volume costume." Your answer must connect to Chapter 1's harmonics (what physically changes when a real instrument is struck harder), name the sampler feature that implements this, and explain why flat velocity 100 therefore wastes money on exactly the libraries that cost the most.

A3. The forum reply. (15 min) A post reads: "Programmed a piano ballad, sounds robotic. Hit Humanize (±20 ms random) and now it sounds like my drunk uncle. Is the plugin broken?" Draft a three-paragraph reply in this chapter's spirit: paragraph one, why random jitter fails (what real players' deviations have that randomness doesn't); paragraph two, the five families of structured humanization, one sentence each; paragraph three, the 20-minute fix list for their specific clip, in order of audible payoff.

A4. Quantize court. ★ (10 min) For each scenario, choose a quantize approach (100% / partial strength / swing / none / manual nudges) and defend it in one sentence of tradeoff language: (a) a hyperpop synth stab pattern; (b) a keys part you played in at half tempo, mostly good, two flubbed beats; (c) a clicked-in pad progression with slow attacks; (d) a bassline that feels right but flams audibly against the (programmed, gridlocked) hats; (e) a chiptune lead.

A5. Sampler or synth? (10 min) For each job, pick sampler, synth, or layer-of-both, and name the tradeoff you're accepting: (a) believable felt piano under an intimate vocal; (b) a bass deeper than any bass guitar for an electronic drop; (c) a string pad that only needs to gesture at "strings" behind a dense mix; (d) a solo cello feature carrying a bridge; (e) a pad that slowly transforms across 16 bars into something no instrument could make.

A6. Why libraries breathe. ★ (10 min) Define multisampling, velocity layers, and round-robins in one sentence each, then answer: which of the three does the machine-gun effect expose, which one does flat velocity waste, and which one does re-pitching chipmunkiness expose? Finish with the two ten-second tests from the chapter that sort deep libraries from shallow ones.

A7. The case for stiff. (10 min) Write a six-sentence defense of 100% quantization as an aesthetic choice, citing at least two genres or artists from the chapter where the grid is the point. Then add the one-sentence caveat that separates intentional stiffness from the doorbell problem.

A8. Idiom audit. (10 min) Each of these programmed parts contains a physical impossibility or idiom violation. Name it and prescribe the re-voicing: (a) a "piano" chord of eleven notes spanning C1–C6, all struck simultaneously; (b) a "bass guitar" sustaining three-note chords under a chorus; (c) a "solo trumpet" holding one unbroken note for 90 seconds; (d) an "acoustic guitar" part where all six chord notes attack within the same millisecond; (e) a "string section" playing staccato chords with zero variation between the twelve repeats.

Part B: 🎧 Listening Lab

Your listening journal (Chapter 4) open, reference level set. The skill being built: hearing the programming decisions inside records — stiff or human, velocity shape, the ghost or the grid. Substitute equivalent tracks from your own library where noted; the named tracks are widely available starting points, not homework you can fail.

B1. The machine as meaning. ★ (20 min) Listen to Kraftwerk's "The Robots," then Stevie Wonder's "Superstition," one full pass each. For Kraftwerk: describe three places where mechanical precision is the aesthetic — what would humanizing destroy? For Wonder: focus on the layered clavinet parts — log every push, drag, and velocity accent you can hear in eight bars (the parts are played, and gloriously uneven). Deliverable: a journal line for each track answering "where does this music keep its life — in deviation, or in relentlessness?"

B2. Hi-hat velocity forensics. (25 min) Hi-hats are where velocity work is most audible on modern records. Listen to Future's "Mask Off" (programmed trap hats: rolls, triplet bursts, velocity ramps) and then D'Angelo's "Untitled (How Does It Feel)" (a band famous for playing behind the beat — loose, human, deliberate). For the trap track: in one 8-bar section, log how the hat rolls change loudness across each burst — flat, ramping up, or ramping down? For D'Angelo: pick any instrument and describe where it sits against the pulse (ahead/on/behind) and whether that placement is consistent. Deliverable: one sentence stating which chapter concept each track demonstrates — structured velocity vs consistent lean. (Full hat language arrives in Chapter 13; today you're just learning to hear it.)

B3. MIDI driving a real piano. (15 min) Aphex Twin's "Avril 14th" — by well-documented accounts, performed via a computer-controlled player piano (a Disklavier): MIDI instructions, real hammers and strings. Listen twice. First pass: would you have known? Log what reads as human (timing? dynamics? pedal?) and anything that reads as subtly mechanical. Second pass: focus only on note lengths and the spaces between phrases. Deliverable: a journal entry on what this track proves about where "realness" lives — the sound source, or the instructions. (It's the player-piano section of the chapter, audible.)

B4. Two pianos, two worlds. ★ (20 min) Alicia Keys' "If I Ain't Got You" (played piano: rolled chords, rubato, phrase-shaped dynamics) versus Robin S's "Show Me Love" (the iconic stabbed house organ/keys riff: gridlocked, velocity-flat, and perfect that way). For Keys: log three specific humanization "moves" from the chapter's list that you can hear her hands making for free. For Robin S: explain in two sentences why the stiffness serves the track — what would partial quantize and velocity arcs do to that riff's job? Deliverable: both logs, plus a one-line answer to "which one is better programmed?" (Trick question. Both match intention to aesthetic.)

B5. Hyperpop chrome. (15 min) 100 gecs' "money machine" (or any SOPHIE production, e.g., "Immaterial"). This is the grid polished into chrome: hyper-quantized, hyper-tuned, velocity-pinned by design. Log: three sonic features that would read as "mistakes" by this chapter's realism rules but are clearly the point here, and one moment where something human-feeling sneaks through the machine anyway. Deliverable: a journal line on what this genre's "contract" (preview of Chapter 18) expects from the grid.

B6. The orchestral gap. (20 min) Listen to two minutes of any John Williams cue recorded with live orchestra (the Jurassic Park theme is a fine choice), then two minutes of a 1990s sequenced game-console score — the PlayStation-era Final Fantasy VII soundtrack ("Aerith's Theme") is the classic example, its orchestral parts rendered by the console's sample playback rather than players. Don't judge the composition — both are beloved for reasons. Log only the performance gap: where do you hear breath, bow, section spread, and phrase-shaped dynamics in the live recording that the sequenced version can't supply? Deliverable: your top-three list of "what fake strings are missing," written before you read Case Study 2 — then check your list against it.

B7. (Optional) Your genre's verdict. (15 min) Pick the most recent track you love in your own genre and run the stiff-or-human audit instrument by instrument: drums, bass, keys/pads, leads. Deliverable: a quick map (instrument → machine/human/hybrid) and one sentence on whether the combination matches what the chapter calls "the machine and the ghost dancing."

Part C: 🎛️ DAW Lab

C1. The stiff-then-humanized drill. ★ (30 min) The chapter's signature exercise. Click in a four-chord progression (one bar each, any key, piano or pad patch): every note on the grid, every velocity at 100, every chord block-simultaneous, lengths exactly full bars. Save/duplicate as Version A. Now spend exactly twenty minutes on Version B — same pitches, nothing added: velocity arcs and top-note hierarchy; chords rolled 5–20 ms; one consistent timing lean somewhere; note lengths that breathe; CC64 pedal (keys) or a CC1 hairpin per chord (pad). Bounce both to audio, close your eyes, alternate. Deliverable: both versions kept, plus three journal lines: what changed most, what you over-did, and the one move with the highest payoff-per-minute. If you can, play both for someone who doesn't make music and write down their words.

C2. Velocity-curve drills. (20 min) Three drills on one melodic patch: (1) program an 8-note scale ascending with velocities forming a perfect arc (e.g., 64-72-80-88-96-88-80-72) — listen for the timbre shift, not just volume; (2) program one note repeating sixteen times, velocities alternating in a hand-motion pattern (e.g., 86/70/78/66...) — then flatten to 100s and compare for the machine-gun cousin; (3) take any chord progression and set every chord's top note +8 over its inner voices, bass note +4. Deliverable: a journal line per drill naming what your specific instrument did with the velocity differences (layer switches? brightness? just level?).

C3. The quantize-strength ladder. ★ (25 min) Record a simple part from a controller (or your computer keyboard's MIDI input) at half tempo — eight bars of chords or a melody, deliberately performed with feel, flubs welcome. Duplicate it four times and quantize the copies at 0% (raw), 50%, 80%, and 100%. Restore full tempo. Listen blind if you can (shuffle the order). Deliverable: rank the four for feel and for tightness, note where they disagree, and write your personal default strength for this kind of part. Bonus: add swing to the 100% copy and log what changes about its attitude.

C4. Build a realistic keys part, start to finish. ★ (45 min) Run the chapter's seven-step workflow as a formal session: cast the instrument (sampled keys), get notes in (played or clicked), partial quantize, velocity pass, length/articulation pass, CC64 pedal lane (down after each chord lands, up at the change), then the subtraction pass — remove at least two notes or one doubling that the part survives without. Deliverable: the part itself, plus your session note documenting the stiff-or-human commitment and three structured deviations you programmed. This part can live in your project — checkpoint work and lab work are allowed to be the same twenty minutes.

C5. The hairpin drill. (20 min) Load any sustained patch (pad or strings). Program one whole-note chord per bar for four bars, velocities moderate and even. First pass: listen to the stare. Second: draw or perform CC1 so every held chord has a shape — swell into bar two, bloom through bar three, decay out of bar four. If your controller has a mod wheel, perform the lane in one take and keep the wobble. Deliverable: before/after bounce, plus one journal line: drawn vs performed — which sounded more alive on your system, and what did the performed lane's imperfections add?

C6. Library depth audit. (20 min) Run the chapter's two tests on three virtual instruments you already own (stock plugins count): (1) one note at velocities 30/60/90/120 — does character change or just loudness? (2) one note repeated eight times fast — machine gun or life? Deliverable: a three-row table (instrument → velocity layers? → round-robins? → verdict), and a note on which instrument deserves your realism jobs versus your proudly-synthetic jobs.

C7. The half-tempo heist. (20 min) Pick a part you "can't play" — too fast, too intricate. Drop the session tempo to half (or less), loop-record it from your controller until one pass feels good, restore tempo, partial-quantize. Deliverable: the part, plus a journal line on what survived the speed change (everything — and write why, in instructions-not-audio terms). If you have no controller: click in the part, then perform only the CC1 lane live with your mouse during playback — even mouse-performed motion beats ruler-drawn geometry.

C8. (Optional) Commit ceremony. (10 min) For every MIDI part now in your project, write the one-line aesthetic commitment in your session notes: "stiff on purpose," "humanized," or "hybrid: ___." Then bounce one finished MIDI part to audio and mute the MIDI original — the anti-fiddling move. Notice your feelings. Log them honestly.

Part D: Synthesis & Critical Thinking

D1. The humanization spec. ★ (30 min) Write a one-page "deviation spec" for your genre, as if briefing a collaborator: which parts ride the grid at 100%, which get partial strength (and how much), where the consistent leans live (what's early, what's late, by roughly how much), velocity ranges per instrument role, and which CC lanes are mandatory on sustained parts. This document is your track's feel, written down — and a draft of the genre conventions work coming in Chapter 18.

D2. The realism budget memo. (20 min) Realism costs time: every layer of humanization is minutes you're not spending on arrangement or sound. Write a half-page memo allocating a strict 60 minutes of programming-polish across a hypothetical 12-track production: which three parts get deep humanization (and why those — think exposure: what the listener actually focuses on), which get the 5-minute pass, and which get gridlocked on purpose. Defend the triage in tradeoff language. (Hint from the mixer's hierarchy you'll meet in Chapter 20: exposed and up-front earns the minutes.)

D3. Stiff-or-human manifesto. (15 min) Revisit your one-page production manifesto from Chapter 5 and add a section: where does your record live on the machine↔human axis, per instrument family? Cite one reference track (Chapter 4 playlist) whose grid philosophy you're borrowing. One paragraph, no hedging — the point of a manifesto is to be wrong on purpose rather than vague by accident.

D4. MIDI 2.0, skeptically. (20 min) The sidebar claims higher resolution "makes structured intention smoother — it doesn't supply the intention." Stress-test that claim: name two concrete situations from this chapter where 7-bit resolution audibly limits you today (think slow CC sweeps, fine velocity gradations on deep libraries), and two amateur problems from the common-mistakes table that 16-bit velocity would not fix at all. Verdict in one sentence: who actually benefits from MIDI 2.0 first?

Part M: Mixed Practice

Interleaving earlier chapters — spaced retrieval is how this sticks.

M1. (Ch 2 + Ch 9) ★ A .mid file has no sample rate, no bit depth, and cannot clip. Explain why all three statements are true — and then name the exact moment in the signal flow where sample rate, bit depth, and clipping start applying to your programmed pad, and which Chapter 2 habit (think headroom at the virtual instrument's output) you should apply when bouncing it to audio.

M2. (Ch 4 + Ch 9) Your new pad sounds gorgeous solo and turns the chorus to mud. Using Chapter 4's band vocabulary and masking concept, explain why a close-voiced chord with roots around 100–200 Hz is the likely culprit, and give two programming fixes from this chapter (no EQ allowed — that tool arrives in Chapter 22) that thin the pileup.

M3. (Ch 6 + Ch 9) You want one shared reverb on the pad, keys, and a future vocal, but the pad also needs its own dedicated delay. Using Chapter 6's routing vocabulary, describe the insert vs send/return layout you'd build, which buses from your template the pad and bass tracks feed, and why the shared reverb as a send saves both CPU and cohesion.

M4. (Ch 8 + Ch 9) You programmed your bass sketch at 1 a.m. on closed-back headphones whose lies list says "lows 60–150 pushed hard." Predict the two most likely errors in the sketch (think level and note-length perception when everything sounds huge), and name the Chapter 8 cross-checks to run before trusting it.

M5. (Ch 3 + Ch 9) Your MIDI controller is plugged in over USB, and you hear a noticeable lag between pressing a key and hearing the pad. Walk the Chapter 3 signal chain to locate where the delay lives (hint: it isn't the MIDI cable — MIDI messages are tiny), connect it to Chapter 2's buffer-size tradeoff, and give the two-setting workflow from Chapter 8 that solves it for tracking vs mixing.