Case Study 1: Three Broken Mixes — The Casebook

The three patients below are composites — assembled from the mixes that land in every working engineer's inbox, year after year, with the names and details invented. The diseases, the checks, and the fixes are real practice. Read each case with the symptom table open, and do the diagnostic work yourself before reading each verdict: the chart comes first in every case, and the chart is enough.

Patient 1: The Muddy Bedroom Beat

The chart. DeShawn is seventeen, makes melodic trap in his bedroom in Cincinnati, and mixes on the bass-heavy consumer headphones he's owned since middle school. His beat — a four-bar loop of chopped soul sample, pad, keys, 808, and hats under a hook vocal — has been slapping on those headphones for two weeks. Then his sister plays it on the kitchen Bluetooth speaker and it comes out as a honking blur. In his mom's car, the low half of the beat is one warm mass; the vocal chop that makes the beat is gone; the kick hits but means nothing. On his headphones, right afterward: warm, full, fine.

The diagnosis. Run the attribution test first: a commercial trap reference on the same kitchen speaker is clear, and in the same car it's tight. The systems are honest; the mix is sick. Symptom vocabulary: thick, congested, blanket over the song, worst everywhere the arrangement is dense, invisible on bass-hyped headphones — that's the muddy row, address 200–400 Hz, textbook to the letter. The reason DeShawn never heard it is the reason the table lists headphones as mud's favorite hiding place: his headphones flatter that whole region as "warmth," so for two weeks he's been listening to the disease and calling it a vibe.

The 60-second check: loop the hook, mute the three biggest low-mid suspects together — pad, soul chop, keys. The blanket lifts. Committee confirmed. Then the solo-reads-the-confession step, one suspect at a time: the soul sample carries baked-in vinyl rumble and a room's worth of low-mids from whatever decade it was recorded in; the pad is a preset with a chest like a furnace; the keys double the pad an octave down for no reason DeShawn can defend; and the 808 — a sampled one, not synthesized — has a smear of 200–300 Hz junk riding above its fundamental.

The treatment, up the ladder. Arrangement: the pad, the chop, and the keys all sustain through all four bars, all song long. First fix is free: the pad sits out of the verses entirely and the keys drop their lower octave — two mutes' worth of decision that removes maybe half the committee before any plugin opens. Source: the 808 sample gets swapped for a cleaner one from the same kit family — the smear was in the file, and no EQ cut removes a sound's character as cleanly as choosing a sound without it. Edit: the soul chops get their tails trimmed so each chop stops ringing into the next — what read as mud was partly the sample overlapping itself. Mix: what's left is small work, exactly as the chapter promised: a high-pass on the chop by ear (the rumble goes; the music stays), wide -2 to -3 dB cuts around 280–320 Hz on the pad and keys, and a high-pass on the reverb return that's been quietly re-depositing low-mids behind everything.

The outcome. Kitchen speaker: the vocal chop is back, the honk is gone. Car: the kick has meaning again, and the 808 is a note instead of a weather system. Headphones: slightly less warm than before — which DeShawn initially reads as loss, until he A/Bs against his reference at matched loudness and hears that the reference carries exactly that much warmth and no more. The lesson he wrote down, which is the lesson of the whole row: his monitoring's flattery was the disease's camouflage, and the committee was convicted by mutes, not meters.

Patient 2: The Harsh Rock Demo

The chart. Erin plays guitar in a four-piece punk band and mixed their three-song demo herself in the bassist's basement. Her note: "It sounded sick for the first week. Now I literally can't listen to a whole song — and my drummer's girlfriend said it 'feels loud even when it's quiet,' whatever that means." On first listen the mix is exciting — bright, aggressive, forward. Ninety seconds in, the cymbals are slicing, the vocal consonants spit, both guitars have turned to wasps, and your hand is on the volume knob. The drummer's girlfriend, it turns out, gave the most precise description in the file: harshness reads as loudness independent of level. The demo demos well and lives badly.

The diagnosis. Fatigue with a 2–6 kHz return address is the harsh row, and the table says to suspect two stacking mechanisms before any single villain. Run the census: across the session there are eleven boosts between 2 and 6 kHz — vocal +3 at 3 kHz "for cut," each guitar +2.5 at 3.5 kHz "for bite," snare +2 at 5 kHz "for crack," overheads bright-shelved "for air" — plus five saturation stages: two pushed amp sims, a saturator on the drum bus, tape emulation on the 2-bus, and a clipper Erin forgot was on the master from a loudness experiment. Every individual move defensible. The sum, a weapon. The confirming check: bypass the 2-bus color and the drum-bus drive together, match loudness, and the glare drops by half before a single EQ is touched.

The treatment, up the ladder. Arrangement: in every chorus, three guitar layers strum the same high voicings in the same register where the vocal needs to live. One of the three re-voices down an octave — a part decision, not a mix decision, and the midrange war de-escalates immediately. Source: both amp sims are running a preset whose character is fizz at exactly 4–6 kHz; twenty minutes of re-picking presets with the mix playing (not in solo, where fizz reads as detail) replaces an hour of corrective EQ. The overheads were also recorded with the mics aimed at cymbal bells in a low concrete room — noted honestly in the log as a next-time prevention item, because a demo doesn't get a re-record. Edit: nothing — this disease didn't live there. Mix: the de-escalation pass. One dB back from each of the three biggest presence boosts rather than three dB from any single one; the clipper comes off entirely (its loudness job belongs to mastering, and it was charging interest in glare); the drum-bus drive backs off a notch with the output re-matched; and a de-esser goes on the overheads — the chapter wasn't joking; hats and cymbals take one beautifully — plus a dynamic EQ on the vocal dipping 2 dB only when the consonants spike.

The outcome. At matched loudness, the treated mix initially sounds duller to Erin — for about thirty seconds, exactly as the table predicts, because brightness reads as quality until it reads as pain. Then the ten-minute test: she plays the whole demo through in the van, all three songs, no volume adjustments, no flinching. The band can finally listen to their own record long enough to like it. Her log's closing line is the row's whole moral: we kept making it brighter because brighter sounded better, and it kept sounding better right up until nobody could stand it.

Patient 3: The Lifeless Singer-Songwriter Ballad

The chart. Grace writes piano ballads and recorded her best one properly: good piano capture, a vocal take with real ache in it, a string pad, brushed drums entering at the second verse. The mix is clean. Balanced. No mud, no harshness, nothing to point at. And it does nothing. She's shown it to four friends; all four said "it's nice" and changed the subject by the second verse. Her question is the hardest one in the casebook precisely because there's no obvious crime scene: "Nothing's wrong with it. So why doesn't anybody care?"

The diagnosis. No frequency address, so the analyzer is useless — this lives on the time axis. Run the lifeless row's three checks. First, GR meters through the verse-to-chorus transition: the vocal compressor sits at 6 dB of reduction everywhere, the piano compressor at 5, the 2-bus glue at 4 — and none of the needles change when the chorus arrives. The chorus is not allowed to get bigger. Second, the bypass test: 2-bus and piano compression off for thirty seconds, matched loudness — the corpse twitches; the piano suddenly leans into phrases the compressor had been politely flattening. Third, the blind energy A/B: loop four bars of verse, then four of chorus, eyes closed. Grace cannot reliably label them. That third result is the deepest one, because it points upstream of every plugin: the strings pad the entire song from bar one, the piano plays the same density throughout, and the drums, once in, never change intensity. The arrangement has no curve for any mix to reveal.

The treatment, up the ladder. Arrangement first, exactly as the row insists, because no automation can resurrect a flat energy map: the strings come out of verse one and the first chorus entirely — their entrance at the bridge becomes an event instead of wallpaper; the piano's left hand simplifies in the verses so the chorus's full voicing has somewhere to arrive from; the brushed drums sit out until the second chorus. Three subtraction decisions, and the song has a shape before a single setting changes. Source: nothing — the performances were never the problem, and the case log says so explicitly, because not every rung gets climbed. Edit: one deliberate non-fix worth recording: Grace had quantize-tightened the piano's rubato in an earlier panic, and the elastic timing goes back in. The drift was the breathing. Mix: halve the gain reduction at every stage — the vocal compressor to 3 dB on peaks, the 2-bus to barely a needle's width — and then the Chapter 27 pass this mix never received: vocal rides phrase by phrase, a +0.5 dB lift into each chorus, the long-tail reverb swelling under the final chorus only.

The outcome. Same notes. Same takes. The next listener Grace plays it for goes quiet during the bridge and asks her to run it back — which is the entire difference between nice and felt, achieved without one new sound. The quiet test now passes: at kitchen volume you can hear the chorus arrive from the next room. Her log entry: the mix wasn't broken; it was embalmed. Every compressor was defensible. The stack was a funeral.

The Pattern Across the Ward

Three patients, three different diseases — and one identical shape underneath. In every case, the top item on the fix list was not a plugin: a pad muted, a guitar re-voiced, a string section benched. In every case, the monitoring or the marketing of the symptom ("warmth," "brightness," "clean") was the camouflage. And in every case the 60-second check ran before the treatment — the mute test, the census, the GR meters — so the fix was aimed at a confirmed diagnosis instead of a feeling.

That's the casebook's quiet argument, and it's the chapter's: mixes don't die of their diseases. They die of unconfirmed diagnoses and panicked treatment. The table exists so that the next time a mix of yours comes back from the kitchen speaker a honking blur, you reach for a chart instead of a fader — and climb the ladder from the cheap end, where most of the cures live.