Appendix D: Audio Reference

The screenshot-and-scan card for production sound — Chapters 14, 15, and 33 compressed into tables you can check on set and at the timeline. It assumes those chapters taught you the why; this is the what and the number. The single idea beneath all of it: sound is half the picture (Chapter 14) — get the right mic close to the source, aim it, reject the rest, and give sound equal weight to light.

Notation used throughout: dBFS (decibels relative to full scale; 0 dBFS = the digital ceiling, where clipping happens), LUFS (Loudness Units relative to Full Scale; a program's perceived loudness over time), dB (a relative change in level), Hz / kHz (frequency). In the diagrams: ((• = microphone, ● = mic capsule, ↑ = the subject in front of the mic, → = sound/signal travelling, ))) = a wireless (radio) link.

The whole appendix in one line. Capture dialogue close, on-axis, out of frame at ~-12 dBFS with headroom below 0 dBFS; monitor on headphones and record room tone every time; in post, clean → EQ → level → music → SFX/ambience → master to loudness. Everything below is detail on those six moves.

D.1 Microphone selection — match the mic to the job

Choose by the job, not the brand. Four families cover almost everything you will shoot.

Lav — wired vs wireless.

The boom is a delivery system, not a mic. A boom is the pole; you mount a shotgun or supercardioid on it to get a directional mic close and out of frame — the backbone of film and TV dialogue. Full booming technique with a partner is Chapter 15. Buy exactly one audio thing for this whole book? Buy a lav — the highest-return purchase in video.

D.2 Polar patterns — where each mic listens

A polar pattern is the map of a mic's directional sensitivity — the invisible shape of what it hears.

FIGURE D.1 — Four polar patterns, seen from above
   (↑ = the subject, in front of the mic;  ● = the capsule)

   OMNI              CARDIOID          SUPERCARDIOID       SHOTGUN (lobar)
   (most lavs)       (many handhelds)  (interior dialogue) (film / outdoors)
        ↑                 ↑                  ↑                    ↑
     ,-----.           ,-----.            .-`   `-.              |''|
    /   ●   \         /   ●   \          /    ●   \             |● |  long,
   (         )       (  front  )        (  front  )            |  |  narrow
    \       /         \       /          \      /              |  |  reach
     `-----`           `--.--`            `-.o.-`               `o `
   hears every       front + sides,      tight front,        tiny REAR lobe
   direction; no     rejects the         small REAR lobe     (o); rejects the
   rejection         REAR (behind ●)     (o) to watch        sides / the room

Off-axis rejection (the whole point): point the sensitive (on-axis) direction at the mouth, and the deaf (off-axis) zone at the noise. For a directional mic, aim is the recording (Chapter 14).

D.3 Mic placement — close, on-axis, out of frame

Almost every dialogue-audio problem is a placement problem, and almost every fix is one of three rules.

FIGURE D.2 — Why closer wins: the direct-to-reverberant ratio
   Halving the mouth-to-mic distance dramatically lifts the direct voice over the room.

   MIC FAR (across the room)              MIC CLOSE (6–8", or a boom a hand off-frame)
   ( S )↑ · · · · · · · · · ((•           ( S )↑((•
        voice + FULL room echo                 the VOICE dominates; the room falls back
        + noise, roughly equal          →      clean, present, "talking TO you"
   = "a recording OF a room"                    = "professional" — and nothing else changed

Two-mic safety: on important dialogue, record a lav and a boom at once — lav = the consistent backup, boom = the natural preferred sound. It covers a rustle, a dead battery, or a drifted aim, and gives the editor a choice (Chapter 14; drilled in the interview, Chapter 19).

On- vs off-camera: get the mic off the camera and onto the subject. When the off-camera mic records to its own device, that is double-system sound — sync it in the edit (Chapter 15; the edit-side sync is Chapter 27).

D.4 Levels and loudness reference

Two different jobs, two different scales. Capture levels (dBFS, on set) protect against clipping; delivery loudness (LUFS, in post) sets how loud the finished program plays back. Do not confuse them.

Capture — set gain by these (Chapters 14–15):

FIGURE D.3 — The dBFS capture meter: aim for -12, leave headroom
    0 dBFS ┤▓▓  ← THE CEILING — never touch it (clipping is permanent)
   -6      ┤██        ← the loudest peaks may flick up here, no higher
   -12     ┤██████    ← AIM: normal dialogue lives here
   -24     ┤███████
   -45     ┤████████  ← room tone / quiet moments
   -60     ┤█         ← the noise floor (system hiss)
   -inf    ┴──────────
   The gap from your peaks up to 0 dBFS is HEADROOM. Too quiet is recoverable; too loud is not.

The asymmetry to memorize: a little too quiet is a nudge in the edit (you pay only in some hiss); a clipped peak is a re-shoot. When in doubt, record a touch low and protect the headroom.

Delivery — master by these (Chapter 33). Streaming platforms normalize loudness, so you master to a target, not past it. LUFS measures how loud the whole program feels; the peak ceiling (dBFS / dBTP true peak) guards against distortion.

Hedge these numbers. The targets above are typical values at the time of writing. Platforms publish and periodically revise their loudness normalization, and targets differ by destination — always confirm the current number at the source before you deliver. The full delivery/codec specs and how to match them live in Appendix H (and Chapter 36). (LKFS is the same measurement as LUFS under another name.)

dBFS vs LUFS in one line: level dialogue by its peaks (dBFS) during the mix; master the finished program by its loudness (LUFS) at the end.

D.5 Location-sound checklist — run it every time

The discipline of Chapter 15, as a checklist. Run it the way a pilot runs a preflight.

Before you roll:

[ ] Headphones ON — closed-back, wired. Monitor the RECORDER'S output, not the room.
[ ] Kill every motor — HVAC, fans, fridge, dimmers, appliances. (Note anything to switch BACK on!)
[ ] Mic close, on-axis, out of frame (D.3). Lav centered & hidden, or boom a hand off the frame.
[ ] Set gain: dialogue ~-12 dBFS; test the LOUDEST line stays under ~-6 dBFS. AGC off.
[ ] Tame the room if it rings: get closer + hang blankets/coats behind the mic AND the subject.
[ ] Wind? foam windscreen (indoor / breath) · furry cover (outdoor) · your own body · a sock.
[ ] Rumble / handling? shock mount + high-pass (low-cut) at ~80 Hz.
[ ] Double-system? camera mic ON as a SCRATCH track; CLAP once, in frame, per take (→ sync, Ch.27).

Monitor for these — the meter shows level, not quality, so your ears are the only judge:

FIGURE D.4 — What to LISTEN for on headphones
  [ ] Distortion / clipping ..... crackle or harsh edge on loud words → lower gain
  [ ] Hum / buzz ................ steady electrical drone (mains, 50/60 Hz) → move cables, kill dimmers
  [ ] HVAC / fridge / fans ...... a low whoosh or motor → turn it OFF
  [ ] Wind / rumble ............. low roar or thumps → wind protection, high-pass
  [ ] Clothing rustle (lav) ..... scratchy friction on movement → re-mount the lav
  [ ] Plosives ................. "p"/"b" pops thumping the mic → angle it off-axis, add foam
  [ ] Echo / room .............. hollow "bathroom" sound → get closer, deaden the room
  [ ] Off-mic / swimming voice .. level rises & falls as they turn → re-aim, brief the subject
  [ ] Intermittent hits ........ a plane, a truck, a phone buzz → pause, re-take
  [ ] Dropouts (wireless) ...... momentary silence / static → batteries, range, interference

Before you wrap — the one habit beginners always skip:

[ ] ROOM TONE: record 30–60 s at EVERY setup — same mic, position, and levels; everyone still & silent.
    It fills gaps, smooths cuts, patches edits, and becomes the noise print + ambience bed in post.
    A borrowed room's tone will NOT match — every location gets its own.

Fix-on-set vs fixable-in-post: reverb and wind roar you must fix here — post cannot cleanly remove them; a steady hum you may leave for noise reduction. But a clipped peak, an echoey room, and un-recorded room tone can never be repaired — fix it in pre and on set, not in post.

D.6 Audio-post order of operations

Lock the picture first — every fix below is keyed to a frame that must not move (Chapter 33).

FIGURE D.5 — The audio-post pipeline: dialogue first, master last
 LOCKED    ORGANIZE   DIALOGUE   EQ &      MUSIC     SFX &      FINAL MIX   MASTER &
 PICTURE → & SYNC   → CLEANUP  → LEVEL   → BED     → AMBIENCE → (balance) → DELIVER
           stems:     de-noise,  cut mud,  license   room-tone  dialogue    ~-14 LUFS,
           DIA/MUS/   de-hum,    add       & duck    bed +      on top;     peak ≤ -1 dBFS;
           SFX/AMB    de-click   presence  under DIA spot SFX   automate    QC on 4 systems
   Dialogue is the boss of the mix: if any layer makes a word harder to hear, that layer is too loud.

The dialogue chain (in order — cleanup, then tone, then dynamics):

Rule: cut problems before you boost, and reduce noise to just enough — overdone, the voice goes "underwater." Bypass the chain often to compare against the raw take.

Stem levels (starting points; the ratios between them are the real lesson):

Music, briefly (Chapter 33; licensing in Chapter 38): match the arc, not just the genre; prefer one evolving theme to a playlist; leave room for the voice; and only ever use music you can prove you're licensed to use (royalty-free / production library, Creative Commons per its exact terms, commissioned original, or a platform's cleared library). Then duck it under the dialogue.

QC before you call it done: play it on four systems (good headphones, computer speakers, a phone speaker, and a car or TV), check it in mono, and watch it through once at final loudness. The phone-speaker pass is non-negotiable. And captions are not optional — they don't replace a clean mix, but a clean mix makes accurate captions (Chapter 36).

D.7 Troubleshooting — symptom → cause → fix

The field-and-timeline diagnostic. Set = fix it on the shoot; Post = fixable at the timeline.

D.8 Numbers to memorize

The short list, in one place.

The one sentence: capture clean dialogue close, on-axis, and out of frame at ~-12 dBFS with headroom, monitor on headphones, and record room tone every time — then clean, EQ, level, score, design, and master to your platform's loudness — because sound is half the picture.