Week 11: Directing with sound
Dialogue and voice performance, ambience, Foley, sound effects and music — how sound redirects emotion, pacing and attention
10 May 2027
A film with picture locked and no sound pass is still a rough cut, no matter how finished the images look. Sound is where the last third of the audience’s attention actually lives.
What you’ll be able to do#
- Direct a voice performance with the same single-note discipline used on a face — a note aimed at pace, breath, or hesitation, not a mood word.
- Treat a location’s sound signature as a second bible entry, fixed and checked for continuity the same way a visual reference is.
- Place a music cue as a directorial statement about how a moment should feel, and hear precisely what it adds by comparing a cut with and without it.
The lecture runs 90 minutes: recap 5, outcomes 5, technique and worked comparison 45, comparison 15, judgement practice 15, synthesis 5.
Directing a voice like a face#
Week 1 directed a line reading with a single note aimed at something concrete and playable — not “sadder,” but a specific change in pace, breath, or where a hesitation lands. That same discipline applies unchanged to a voice generated on its own, independent of the face it will eventually sync to. “More emotional” fails as a voice note for the same reason it failed as a performance note in week 1: it names a mood, not something a take can be judged against.
Directing a line reading also rewards a kind of responsiveness Sanford Meisner’s actor training names as “living truthfully under imaginary circumstances” — treating the moment as something being lived through in real time, rather than a pre-decided delivery played back. Framing that as a technique specifically for directors is this course’s own inference, not documented Meisner-for-directors pedagogy — no Meisner-authored or Institute-published curriculum for directors was found in preparing this course. What transfers cleanly, regardless of that framing’s provenance, is the practical note: direct the moment being lived, not the line being recited.
Ambience and Foley as a location’s second bible#
Week 5 established that a location bible entry has to be checked, not remembered from scene to scene, because nothing about a generated location persists on its own. Sound has the identical problem. What a kitchen at night sounds like — the low hum of an empty room, a tap dripping, traffic outside — is as fixed a fact about that location as its layout or its lighting, and it needs to be locked and re-supplied the same way, or a sequence’s ambience will drift exactly as its visuals would without a bible to check against.
Music as a directorial statement, not wallpaper#
A music cue laid under a whole scene by default isn’t a directorial decision — it’s the absence of one. Placed on purpose, a cue is a specific claim about how a moment should feel, the same category of choice as a cut point in week 10 or a shot’s argument in week 4 and week 8. Removing the cue doesn’t just make the moment quieter — it removes a claim the scene was making, and the difference is worth hearing directly rather than assuming.
A real voice pair and sound comparison#
These actual sound edits use Week 10’s 31.23-second Cut A of “Before It Stops.” The owner approved applying the primary version below. The picture is identical across all versions; this teaching choice does not select the final-film cut.
Direct the same line at two paces#
The shared circumstance is that Theo’s call interrupts Nadia’s final packing task in her mother’s kitchen. The line is “Hey.” Two independent macOS Samantha syntheses vary only delivery pace: A at 150 words per minute, B at 105. These are synthesized voice candidates, with no added-silence second take or claimed human performance recording.
Take A — 0.361 seconds
Take B — 0.397 seconds
The primary comparison uses A: its shorter delivery leaves more time for listening after the greeting in the existing ending. This is the editorial reason for the applied choice; no independent listening-based performance selection or audience response is claimed. The slower-voice alternate remains available for comparison.
Dialogue-only#
Full sound pass#
Both versions omit the original generated audio, avoiding duplicate speech, ringing or music. The full pass adds original procedurally synthesized kitchen hum, an incoming-call SFX motif and a sustained A3/E4 music cue. The hum is a proposed sound signature for this teaching mix, not a verified location recording or an approved production bible entry.
| Event | Placement in both comparison pictures |
|---|---|
| Incoming-call SFX | Begins at 6.067 s; stops at 24.1 s near the answer interaction |
| Sustained music cue | Enters at 17.6 s near the visible reach, then fades at the end |
| “Hey” | Begins at 27.8 s, after the phone reaches her ear around 27.7 s |
Isolate what the music adds#
The dialogue-only/full-pass pair changes several sound layers together. Use this control to compare the music alone: it keeps the full pass’s identical dialogue, ambience and SFX, removing only the cue.
Watch the versions before opening the intended comparison below. What changes in your read of each beat, and which sound is responsible?
Compare the intended beat-by-beat changes
| Beat | Intended contribution of the full pass |
|---|---|
| Wide / waiting | A continuous low hum proposes a still kitchen |
| Caller insert | The phone motif makes the interruption audible |
| Reaction | Repeated phone sounds sustain pressure while the picture holds |
| Reach / hover | Music enters with attempted contact, drawing attention to the choice |
| Acceptance / “Hey” | Ringing ceases near the answer; ambience and cue continue beneath the same greeting |
These are editorial intentions to test against what you actually hear, not measured emotional effects. Compare the no-music control with the full pass before attributing a difference specifically to music.
Read the comparison, provenance and verification limits.
Verification: all four videos decode successfully, have identical copied picture streams and run 31.233 seconds; decoded audio peaks stay below clipping. Visual ending samples support event-aligned dialogue placement. The generated mouth remains open across several frames, so exact phoneme lip sync has not been established. Independent listening, audience review and an end-to-end workshop trial remain unverified. These teaching assets do not establish individual assessment completion.
Misconception contrast#
| Sound as a finishing layer | Sound as a directed layer |
|---|---|
| A line note aimed at “more emotional” | A note aimed at pace, breath, or hesitation |
| Ambience however the room happens to render | A location’s sound signature locked like its visual bible |
| Music placed under the whole scene by default | A cue placed on purpose, with a stated claim |
| “It feels more finished” as the only judgement | Naming what changed emotionally, beat by beat |
Guided judgement: what is this cue actually doing?#
A student adds a swelling string cue under the entire closing shot, starting the instant the phone lights up and continuing through “Hey.” Identify what claim that placement makes, and whether it matches the beat’s intent from week 7’s moment-before.
Compare your diagnosis
Starting the cue the instant the phone lights up claims the anticipation is the emotional center of the beat — the swell arrives well before the directed hesitation from week 7 does any work, which risks announcing the outcome before her performance has earned it. If the beat’s intent is that her hesitation, grounded in the call interrupting her final packing task in her mother’s kitchen, is what the moment is actually about, a cue entering later — under the reach itself, not the first ring — matches that intent more precisely. The fix isn’t to remove the cue; it’s to move where its claim starts.
For the workshop, apply the same test to your own cue placement: what claim does the moment it starts on make, and does that match what the beat is actually about?
What you take to the workshop#
Bring your chosen cut from week 10, with one line still undirected.
You leave with two directed line takes and a dialogue-only cut sitting next to a full sound-pass cut — naming, beat by beat, exactly what the pass changed.
Teaching team
- Marisol Quaye — convenor