Step 6 Audio Generation Spec — “The Ferret Incident”

Author: rho-editor | Date: 2026-05-20

Overview

This document defines all audio assets needed for Step 6 (Soundstage). Three categories: Score (6 movements), SFX (9 sound events), and Ambience (1 continuous bed).

SCORE — 6 Movements

All score should be orchestral/chamber music in a whimsical, precise style. Think Desplat scoring Anderson. No electronic elements.

#	Movement	Timeline	Duration	Tempo	Instruments	Notes
1	Clockwork	0s-47s (Sc 0+1)	~47s	80 BPM	Glockenspiel, celesta, pizzicato strings	Precise, metronomic. Opening theme. Must feel mechanical and controlled.
2	Discovery	46s-67s (Sc 2, 2.1-2.5)	~21s	80 BPM	Strips down: solo glockenspiel fading out	Near-silence. Tension building. Score nearly disappears.
3	Escalation	67s-140s (Sc 2, 2.6-2.21)	~73s	100 BPM	Plucked cello, muted trumpet, pizzicato	Quick staccato phrases. Builds through the chase/cleanup. Gets more frantic.
4	Chandelier	140s-178s (Sc 3)	~38s	70->STOP	Full strings build, then SILENCE	Score builds to crescendo, then CUTS DEAD at 3.4-3.5 (crystal drop implied). 2s dead silence before next movement.
5	Inspector	178s-233s (Sc 4)	~55s	60 BPM	Sparse icy strings, isolated notes	Cold, clinical. Let pen scratch SFX carry rhythm. Score barely present.
6	Return	233s-249s (Sc 5)	~16s	80 BPM	Opening motif returns (glockenspiel + celesta)	Nostalgic callback to Movement I. Resolves with final bell ding. Fade with picture.

Score Mixing

Track volume: -2 dB
Duck under voice: -12 dB (both VO and DLG tracks)
Movements should have clean starts/ends with 0.5s fades at boundaries

SFX — 9 Sound Events

#	Sound	Shot(s)	Timeline Position	Duration	Priority	Notes
1	Bell ding x3	1.3	20s-29s (evenly spaced at ~20s, 23s, 26s)	~0.5s each	HIGH	Silver desk bell. Clear, bright ding. Same bell sound for all.
2	Fern crash	2.3->2.4	~57s-58s (end of 2.3 / start of 2.4)	~1.5s	MED	Off-screen: ceramic pot tipping, dirt spill, fern thud. Muffled.
3	Clock tick (ambient)	1.1, 2.5, 2.18, 3.7	Multiple ECU inserts	Fills each shot	MED	Steady, mechanical tick-tock. Only on clock ECU shots. Not continuous.
4	Suitcase slam	2.12	~97s (2s into shot)	~1s	MED	Leather suitcase being slammed shut. Decisive.
5	Bell slap x1	2.15	~111s (~1.5s into shot)	~0.5s	HIGH	Ferret hitting bell. Slightly different from proper ding — more of a chaotic slap.
6	Crystal drop + shatter	3.4->3.5	~162s-164s (end of 3.4, start of 3.5)	~2s	HIGH	Crystal pendant falling + shattering on marble. The moment score goes silent.
7	Pen scratch	4.4	205s-209s	~4s	MED	Pen on paper/clipboard. Steady scratching throughout Vance writing.
8	Door open/close	4.1, 4.8	178s (open), 233s (close)	~1.5s each	LOW	Heavy brass hotel doors. Weighty, slow creak + click.
9	Final bell ding x1	5.2	~245s (6s into shot)	~0.5s	CRITICAL	THE payoff. Same bell as #1 but single, definitive ding. Last sound before fade to black.

SFX Mixing

Track volume: -3 dB
No ducking — SFX plays at consistent level
Bell sounds must be consistent across all uses (same sample)

AMBIENCE — 1 Continuous Bed

Sound	Scenes	Duration	Notes
Hotel lobby ambience	All (0-5)	Full film ~249s	Very subtle: distant air conditioning hum, faint marble echo. VERY low in mix.

Ambience Mixing

Track volume: -18 dB
No ducking — always present but barely perceptible
Provides a “live room” feel to prevent dead silence between events

Generation Priority

Bell ding (used 5 times across film — the iconic sound)
Crystal shatter (dramatic pivot point)
Score Movement I: Clockwork (sets the tone)
Score Movement VI: Return (emotional payoff)
All remaining score movements
Remaining SFX
Ambience (can be generic hotel lobby)

Timeline Track Additions Needed

When assets arrive, add to timeline.json:

SCORE track: Already has placeholder (track index 3, role: “music”, vol: -2dB, duck: -12dB)
SFX track: Already has placeholder (track index 4, role: “sfx”, vol: -3dB)
AMB track: NEW — needs to be added (role: “sfx”, vol: -18dB)

Audio Specification