feat(audio): NPC conversation murmur asset (SFX-003)

Produces sfx_npc_murmur.ogg via Stable Audio Open for D-072 universal
event-driven NPC proximity murmur. Organic sonic family (D-074): warm
mid-freq 200-800Hz vocal texture, human voices in adjacent room with
direction and cadence distinguishable, words not. World SFX bus.

Generated 3 candidates (5s/CFG8, 6s/CFG7, 4s/CFG9), selected candidate
1 for richest texture and clearest spatial framing. Unblocks client #533.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
2026-02-19 17:10:17 +01:00
co-authored by Claude Sonnet 4.6
parent 61c452aca4
commit a394ff6a5f
2 changed files with 12 additions and 1 deletions
Binary file not shown.
+12 -1
View File
@@ -15,6 +15,7 @@ Mixed methods depending on duration and character:
|----|----------|--------|-----|--------|----------|-------------|--------|
| SFX-001 | `sfx_footstep_metal_walk.ogg` | done | Player Actions | SAO hybrid | 0.3s | Single footstep on metal grating. Walk pace. SAO 5s/100 steps/CFG 8, best transient trimmed at 2.42s, LUFS -16 normalized. | S10 #327 |
| SFX-002 | `sfx_footstep_metal_run.ogg` | done | Player Actions | SAO hybrid | 0.3s | Faster footstep, sprint pace. Louder — player is broadcasting position (D-053 acoustic footprint). SAO 5s/100 steps/CFG 8, best transient trimmed at 1.03s, LUFS -16 normalized. | S10 #327 |
| SFX-003 | `sfx_npc_murmur.ogg` | done | World SFX | SAO | 5s | NPC proximity conversation murmur. Human voices in adjacent room, direction and cadence distinguishable, words not. Organic family. SAO 5s/100 steps/CFG 8, LUFS -16 normalized. Selected from 3 candidates (richest texture). | S13 #532 |
## Detailed Entries
@@ -25,6 +26,17 @@ Mixed methods depending on duration and character:
- **Integration:** Triggered by player movement tick. Positional (AudioStreamPlayer2D at player position) — maps to camera-relative stereo per D-018.
- **Cross-reference:** D-038 (#5), D-053 (movement stances as acoustic decisions)
### SFX-003: sfx_npc_murmur
- **Sonic family:** organic (physical world — human warmth, natural decay)
- **Bus rationale:** World SFX — this is information about other people in the world, not player feedback. Positional (AudioStreamPlayer2D at NPC cluster position).
- **Design intent (D-072):** Universal asset — single murmur plays at all NPC proximity events. Zone ambient determines conspicuousness naturally: bar's dense ambient buries it, corridor's sparse ambient exposes it. No zone-specific variants needed.
- **Aesthetic (D-074):** Organic family. Warm, lived-in, no lattice processing. Mid-freq 200-800Hz texture per station palette "functional warmth" (D-074 conversation murmur band).
- **Propagation (D-018):** Close range: intelligible as "voices, nearby." Medium range: muffled. Long range: lost in ambient floor.
- **Eavesdrop interaction (D-071):** Player ListeningFocus boost increases effective reception range. Same asset, confidence varies with proximity and skill.
- **Generation:** SAO 5s / 100 steps / CFG 8. Prompt: "Warm organic audio, natural sound recording, slight room ambience, lived-in industrial space, realistic acoustic properties, authentic environmental sound. Single isolated sound effect, clean recording, 5 seconds duration, clear attack and natural decay, no background noise, no reverb tail beyond natural. Human voices murmuring through a thin wall or partition, two or three people in low conversation in the adjacent room, words muffled and indistinct, direction and gentle cadence audible, warm mid-frequency vocal texture 200-800Hz, soft intimate conversation murmur, no music, no intelligible words." 3 candidates generated (5s/CFG8, 6s/CFG7, 4s/CFG9). Candidate 1 selected (richest OGG texture, clearest spatial framing). Post-processed: trim, LUFS -16, OGG.
- **Integration:** Event-driven — fires when player enters NPC cluster proximity range. Not looped continuously; retriggered on proximity events. Loop-capable (5s has natural loop structure) if client opts to sustain during eavesdrop.
- **Cross-reference:** D-072 (universal murmur decision), D-071 (ListeningFocus eavesdrop), D-074 (organic aesthetic), D-018 (sound propagation model)
## Future Expansion
SFX will grow significantly with:
@@ -32,7 +44,6 @@ SFX will grow significantly with:
- Door open/close, machinery activation → World SFX bus
- Combat impacts, weapon sounds → Player Actions bus
- Environmental events per planet/biome
- NPC conversation murmur (event-driven, point-source) → World SFX bus
When this file exceeds ~50 entries, split by domain:
- `sfx-movement.md` (footsteps, stance variants)