Case Study 2: A Location-Sound Shoot-Along — Recording a Talk in a Bad Room
Case Study 1 took apart a master's philosophy of location sound. This one builds something from scratch in the least glamorous, most common situation you will actually face: a real event, in a real room, that sounds terrible — and you have one take. Follow the thinking at every step, because the decisions are what you will re-make on your own shoots. By the end you will have recorded a clean, usable talk in a hard, droning room using a boom, a lav, and your ears, and you will have cut a short clip from it.
This is a production-heavy walkthrough, deliberately different in kind from the analysis in Case Study 1. It is set in the book's fourth recurring setup — the event / room in available conditions — and everything here is a constructed teaching example: an illustrative shoot you can reproduce at any talk, workshop, meeting, or ceremony you can get access to.
The brief and the constraints
The brief: a guest is giving a 10-minute talk to a small audience — maybe thirty people — in a community-center multipurpose room. We want a clean recording of the talk we can cut into a 60–90-second highlight, plus enough of the room's life (applause, a laugh, the Q&A) to make it feel like a real event. It has to sound like a professional captured it, and — this is the hard part — there is exactly one take. The talk happens once. Nothing here can be re-shot.
That single constraint changes everything. On a controlled talking-head you can roll again. At an event you cannot, so the entire job shifts toward insurance: redundant mics, conservative levels with plenty of headroom, and relentless monitoring, because the mistake you don't catch live is a mistake you keep forever.
The constraints, stated up front (this is pre-production thinking, which you'll formalize in Chapter 16):
- Gear: one camera on a tripod (a phone is fine); one lav (wired or wireless) for the speaker; one shotgun on a boom (a real pole, or a stand with a clamp for solo work); closed-back headphones; a recorder or a second phone; and, for taming, whatever soft material the room offers. That's the kit.
- The room: a rectangular hall with a hard tile floor, painted cinderblock walls, a low acoustic-tile ceiling, a wall-mounted HVAC unit that drones, and a projector whose fan whirs near the front. It rings when you clap. It is, acoustically, a bad room — which is exactly why it's worth practicing in.
- Sound to capture: the speaker's talk (the priority), the audience's reactions (applause, laughter — the "event" texture), a few Q&A questions from the floor, and room tone.
- One take. Plan for redundancy and headroom accordingly.
🎒 Gear Note: the event insurance kit, itemized. The philosophy at an event is two paths to every sound. The lav on the speaker is your primary, close, consistent track — it rides with them if they pace. The boom (or a shotgun on a stand aimed at the lectern) is your safety and your natural-sounding room presence — if the lav rustles or its battery dies, the boom saves the talk. The camera scratch is your third net and your sync reference. None of this needs to be expensive: a wired lav into a second phone, a shotgun clamped to a light stand, and $15 closed earbuds will do the whole job. What you're buying with redundancy isn't quality — it's survival of the one take.
The gear and the settings
Because there's one take and unpredictable loud moments (applause, laughter, a dropped mic), the settings lean conservative: lower average level, bigger headroom, safety track on, monitoring non-stop. These are starting points to adjust on the day, not a recipe.
⚙️ Settings Box — event / bad-room dialogue capture.
Setting Starting point Why Primary mic Lav on the speaker, close under the collar Consistent, close capture that rides with them (Ch.14, §14.4). Safety mic Shotgun on boom/stand, aimed at the lectern Redundancy + natural room presence if the lav fails. System Double-system (lav + shotgun to a recorder), camera mic as scratch Independence and backup; clap once to sync (§15.1). Dialogue average ~-14 to -12 dBFS(a touch conservative)Extra headroom because applause/laughter spike hard and can't be re-taken. Peak ceiling under -6 dBFS; test with a loud clapProtects against the unrepeatable loud surprise (§15.4). Safety / backup track ON, ~-12 dBlowerThe event insurance policy — a clipped applause peak survives on the safety. AGC OFF No pumping of the room's constant drone in pauses (§15.4). High-pass filter ON (~ 80 Hz)Rolls off the HVAC rumble and floor-borne footfalls (§15.5). Monitoring Closed-back headphones, on the whole time The one take demands you catch every fault live (§15.3). Room tone 60 s, grabbed before the room clears Every event room needs its own (§15.6).
Notice the choices that differ from a controlled interview: the average is a little lower and the safety track is on, both because applause and laughter are the loudest and least predictable sounds in the room and you cannot re-record them. At an event, headroom is not a nicety; it is the difference between a clip you keep and a clip that crackles.
Setup: read the room, then place everything
Walk in early — before the audience — and do the two things that matter most, in order: listen, then place.
First, listen. Clap once, hard, in the empty room and hear the tail — this hall rings for a beat, off the tile and the cinderblock. Walk to the HVAC unit and the projector and listen through your headphones with the shotgun: both drone. You can't rebuild the room and you can't stop an event's HVAC (people need to breathe), but you can make three decisions that stack the deck:
- Get the mics close. Proximity is your reverb weapon (§15.5). The lav rides inches from the mouth; the boom/shotgun sits as close to the lectern as the frame allows. Close mics mean the ringing room and the drone lose to the voice.
- Aim rejection at the noise. Point the shotgun's dead sides toward the HVAC and projector so its pattern rejects them (Ch.14, §14.3). A small rotation of the stand can drop a drone noticeably.
- Flip the high-pass on. The HVAC's energy is mostly low rumble; the low-cut (§15.5) sheds it while leaving the voice intact.
Now place everything. Here is the sound plan, top-down:
FIGURE CS2.1 — Event sound plan: lav primary + shotgun safety in a bad room (top-down)
[ HVAC drone ] ))) ((( [ projector fan ]
| <- shotgun's DEAD side aimed here | <- and here
| |
================= LECTERN =================
( S ) speaker ● (lav under collar, close)
((• shotgun on a stand, aimed at the mouth,
as close as the wide frame allows
window / house light [ CAM ] wide of speaker + lectern
☀ | (camera mic = scratch)
v
[ audience: 30 people — applause, laughter, Q&A ] [ RECORDER: lav ch.1 + shotgun ch.2
+ safety track -12 dB ]
Two paths to the voice; rejection aimed at the two drones; low-cut on; monitor throughout.
The logic of the map: the lav is the star — close, consistent, immune to the speaker turning their head. The shotgun is the understudy and the atmosphere — if the lav rustles against a lanyard or dies, the shotgun carries the talk, and even when the lav is perfect, the shotgun's slightly more open, roomier sound is lovely to blend under it. The camera records a wide of the whole lectern (so we have picture that always works) with its mic as scratch. And the shotgun's dead sides are turned deliberately toward the HVAC and the projector, so the two drones fall into its zones of rejection.
🔗 Connection. Many real events hand you a shortcut this brief deliberately skips: a PA system or sound board already miking the speaker, from which you can take a "board feed" — a direct line out of the venue's mixer straight into your recorder. It is often the cleanest capture available and it is standard practice for conferences and weddings. We cover the board feed properly in Chapter 24 (§24.6), where events get their own chapter. For now, know it exists — and know that even when you take a board feed, you still put your own mic up as a backup and still grab room tone, because a board feed can fail or sound thin, and it never captures the room's life.
The shoot, in phases
An event unspools in a fixed order you don't control, so you prepare for each phase before it arrives.
Phase 1 — Before the doors: levels, sync, and a listen-back
With the room still empty, wire the speaker (or a stand-in), set gain, and test the loudest thing. Have them talk normally — set the lav to average around -14 dBFS — then have them laugh and clap right at the mic, and confirm the peak stays under -6. Turn on the safety track. Then roll everything, clap once in front of the camera, say "room check, take one," and record twenty seconds. Play it back on the headphones. Is the voice clean? Is the HVAC drone acceptably low under the high-pass? Did all three recordings actually arm? This listen-back, done while you still have time to fix things, is the single highest-value two minutes of the day.
FIGURE CS2.2 — The event levels plan: conservative average, big headroom (one channel)
0 dBFS |##| <- applause/laughter must NOT reach here. One take — no re-record. ^
-6 |##| _ peaks (a laugh, a clap) may flick to here, no higher | HEADROOM
-9 |####| __| | (bigger
-14 |#########| <- TARGET: the talk averages here (lower than a studio -12) | than
-20 |#####| | usual)
-40 |##| <- room tone / HVAC drone (pushed down by the high-pass) v
-60 |#| <- noise floor
+--------------------------------------------
At an event you buy MORE headroom than an interview, because the loudest sounds
(applause, laughter) are unrepeatable and arrive without warning.
Phase 2 — The talk: monitor, don't fiddle
The audience files in (their bodies, usefully, deaden the room a little — thirty people absorb sound the empty hall didn't). The speaker begins. Your job now is almost entirely listening. Levels are set; resist the urge to ride them on a talk — a steady speaker wants a steady level, and fiddling introduces jumps. Keep the headphones on and hunt for the faults from the §15.3 checklist: the lav brushing the lanyard when they gesture, a phone buzzing in the front row, the projector fan creeping up as the room warms, the moment they step back from the lectern and the shotgun goes distant.
FIGURE CS2.3 — "The laugh that tests your headroom" [constructed teaching example]
THE FRAME A wide of the speaker at the lectern, audience shoulders dark in the foreground; the room
reads honest and plain — a real community hall, not a studio.
THE MOVE Locked off on the tripod. The camera is stable and patient; we're here for the words.
THE LIGHT The room's own overhead light plus a window (Chapter 13) — available, unglamorous, true to
the event.
THE SOUND The lav carries the talk close and clean; the HVAC sits low and unbothersome under the
high-pass. Then a joke lands and the whole room laughs and claps — a sudden spike. On the
headphones you hear it surge; on the meter it leaps toward the ceiling.
THE CUT This is the moment the highlight will be built around; the laugh is the payoff we'll cut to.
THE EFFECT Because you set a conservative average and kept headroom, the laugh peaks near -6 dBFS and
stays clean — it reads as warm, real, and present, not as a crackling blowout.
THE LESSON At an event, the loudest moment is the whole point AND the biggest risk. Headroom is what
lets you keep it.
What you adjusted: midway through, you heard on the cans that the lav had started brushing the speaker's badge lanyard whenever they turned — a faint scratch on every gesture. You couldn't stop the talk. But because the shotgun was up as a safety, you knew the talk was still being captured cleanly on channel two, and you noted the timecode of the worst rustles so the editor can swap to the shotgun there. That is redundancy doing its job: a fault you can't fix live doesn't sink you if you built a second path to the sound.
⚠️ Common Mistake: taking the headphones off once it "sounds fine." The event is going smoothly, so you relax, set the cans down, and watch through the viewfinder. That is exactly when the lav battery dies, or the speaker walks to the whiteboard and off the shotgun's axis, or feedback starts building in the PA — and you catch none of it until the edit. At an event, the headphones stay on from first word to last applause. The one take gives you no second chance to hear what you missed.
Phase 3 — The Q&A: chasing sound you don't control
Questions from the floor are the hardest event sound: they come from unknown directions, unmiked, from people who mumble. Three moves salvage them. If you have a second person, they can carry a mic to questioners. Solo, you swing the shotgun toward each questioner as they start (favor the talker, §15.2) — you'll get them roomier than the speaker but usable. And the reliable fallback: the speaker repeats the question before answering (a good host does this anyway), so even if you miss the questioner, the speaker's clean lav gives you the substance. Brief the speaker on this beforehand; it's the single best Q&A insurance there is.
⚠️ Common Mistake: the dead wireless battery mid-talk. Events are where wireless mics fail, because they run longest and you can't pause to swap a battery on a live speaker. Two habits prevent the heartbreak: put fresh batteries in every wireless transmitter and receiver right before the talk (not "they were fine yesterday"), and keep the wired shotgun safety up precisely so a mid-talk dropout doesn't cost you the sentence. On the headphones, a wireless dropout sounds like a momentary silence or a burst of static — if you hear it repeating, you're losing range or fighting interference, and the shotgun is now your primary. This is redundancy earning its keep: at an event, every important sound wants two independent paths to your recorder, because the one that fails will always be the one carrying the best line.
Phase 4 — Before the room clears: room tone
The talk ends, the applause fades, people start to stand — and here is the discipline most crews forget in the rush to pack up. Before the room fills with packing noise and empties of its audience, you grab room tone. Ideally you catch a beat right at the end while people are still settling, or you ask for fifteen seconds of quiet. Record 60 seconds of this room — the HVAC, the projector, the murmur — with the same mic and levels. This is the ambient bed that will make your highlight feel continuous when you cut it from a dozen moments.
FIGURE CS2.4 — "Room tone at an event" [constructed teaching example]
THE FRAME The wide, now with the speaker stepping away and the audience beginning to shift — or a held
empty shot of the lectern and hall.
THE MOVE Locked off, unchanged.
THE LIGHT Unchanged — you don't restrike anything; you just keep rolling.
THE SOUND The point of the shot. No talk — just the room's own voice: the HVAC's low breath, the
projector's whir, the soft rustle of an audience not yet gone. Recorded on the same lav/
shotgun at the same level as the talk, so it matches perfectly.
THE CUT Never seen by the audience; it lives *under* the whole highlight in the edit, filling every
gap and smoothing every cut.
THE EFFECT In the finished piece, the event sounds like one continuous place, even though it was cut
from many moments — because this invisible layer holds it together.
THE LESSON The busier and more unrepeatable the event, the more you need room tone — grab it before the
room changes, or it's gone with the crowd.
The edit pass
Back at the timeline (the full craft is Chapters 26–30; here's the sound logic). First, sync: drop the camera wide and the recorder's tracks onto the timeline and line up the clap spikes from Phase 1 (§15.1) — or let the software auto-sync by waveform (Chapter 27, §27.4). Now picture, lav, and shotgun all lock together.
Then build the highlight. The lav is your spine — clean, close, consistent. Where you noted the lanyard rustle, cut to the shotgun for those lines; its slightly roomier tone blends fine and it's clean where the lav wasn't. Lay the room tone across the entire clip as a continuous bed so nothing ever drops to dead silence. Keep the applause and laughter — they're the event's warmth — trusting the headroom you protected to have kept them clean. Here's the sound side of the timeline:
FIGURE CS2.5 — The event highlight: cutting the sound (timeline)
V1 [ wide: speaker ##### ][ CU-ish push ####### ][ audience laugh ### ][ wide ##### ]
A1 [ LAV (spine) ######## | ###### ][ shotgun ##### ][ LAV ############ ] <- swap to
A2 [ shotgun (blend/safety) ..................................... ] shotgun on
A3 [ ROOM TONE bed ............................................... ] the rustle
^ where the lav rustled, the cut on A1 jumps to the clean shotgun
A1 is the primary voice; A2 blends the shotgun under it for room; A3 is the continuous
tone that hides every seam. The laugh (V1/A1) survives clean because you kept headroom.
The key decisions, named: lav as spine, shotgun as rescue and blend, room tone as glue, headroom as the thing that saved the laugh. Every one of those was set up on the day by the disciplines of this chapter — the edit only spent what the shoot banked. That is you shoot for the edit, proven at an event where you get one chance to bank it.
✂️ In the Edit. The single most valuable thing you did for this edit happened in Phase 2, when you noted the timecode of the lav rustles instead of just wincing. A note like "lav rustle 04:12–04:20, use shotgun" turns a fifteen-minute hunt into a fifteen-second fix. Production and post are one craft: the recordist who takes sound notes is the editor's best friend, and at an event — where you can't re-shoot — those notes are sometimes the whole difference between a usable clip and a lost one.
Discussion questions
- The whole shoot is organized around one constraint: there is one take. List three specific settings or choices in this walkthrough that would be different — looser — if you could re-record, and say why the one-take rule tightened each.
- The lav is the primary and the shotgun is the safety. Argue the reverse case: when might you make the shotgun the primary at an event and the lav the backup? What about the room or the speaker would drive that?
- Room tone at an event has to be grabbed "before the room changes." Why is event room tone harder to capture than a controlled-interview room tone — and what do you lose if you skip it?
- The Q&A section offered three fallbacks (a second person with a mic; swinging the shotgun; the speaker repeating the question). Rank them by reliability for a solo shooter, and explain your top choice.
- This room "rings" and "drones," and you couldn't rebuild it. Which of §15.5's tools actually helped here (proximity, absorption, rejection-aiming, high-pass) and which were unavailable at an event — and what does that teach about the limits of taming a room you don't own?
- The 🔗 Connection notes that even with a venue board feed you'd still put up your own mic and grab room tone. Why is "the venue is handling sound" never a reason to bring no sound plan of your own?
Your turn: record a real talk
Take this to an actual event you can access — a lecture, a sermon, a workshop, a wedding toast, a club meeting, a friend's presentation.
The brief: capture a 5–10 minute talk in a real room with two paths to the voice (a lav and a boom/shotgun, or a board feed plus your own mic), monitored throughout, and cut a 60–90-second highlight with a continuous room-tone bed.
Constraints and guidance:
- Scout and listen first. Arrive early, clap in the empty room, find the drones (HVAC, projector, fridge), and aim your mic's rejection at them.
- Set conservative levels with big headroom. Average around
-14 dBFS; test a loud clap stays under-6. Turn the safety track on. Applause and laughter are unrepeatable — protect them. - Two mics, one take. Lav as primary spine, shotgun as safety and room. Clap once to sync. Leave the camera mic on as scratch.
- Monitor from first word to last applause. Headphones on the whole time. Note the timecode of any fault you can't fix live.
- Grab room tone before the room clears. Sixty seconds of the space, same mic and level. Do not pack up without it.
- Cut for the moment. Build the highlight on the clean lav, swap to the shotgun where the lav failed, lay the tone underneath, and keep the laugh.
Play the finished clip for someone who wasn't there and ask two questions: "Could you hear every word?" and "Did it feel like a real event?" If the answer to both is yes, you have done the hardest thing in location sound — you captured a room you didn't control, on a take you couldn't repeat, and made it sound like you were never worried at all.
Key takeaways
- At an event, the whole game is surviving the one take — which means redundancy (two paths to every sound), conservative levels with extra headroom, and non-stop monitoring.
- Lav as spine, shotgun as safety and room. Two independent captures mean a rustle, a dead battery, or a distant moment on one is rescued by the other.
- Aim rejection at the drone, get close, flip the high-pass. You can't rebuild a bad room you don't own, but proximity and a shotgun's pattern turned away from the HVAC and projector win the voice back.
- Protect the loud, unrepeatable moments. Applause and laughter are the event's warmth and its biggest clipping risk; the headroom you keep is what lets you keep them clean.
- Take sound notes. Logging the timecode of a fault you can't fix live turns an edit-time hunt into a swap — the recordist's gift to the editor (usually you).
- Grab room tone before the room changes, and lay it under the whole highlight so a piece cut from many moments sounds like one continuous place.
- Every rescue in the edit was banked on the day. At an event you get one chance to bank it — which is why the disciplines of this chapter are not optional there. They're the job.