fix: align audiovisual and narrative feedback

This commit is contained in:
2026-07-23 23:20:50 +09:00
parent 8cf0886d7d
commit e06b15ce7a
27 changed files with 1368 additions and 142 deletions

View File

@@ -45,6 +45,8 @@ The source files in this section were downloaded from their official Pixabay det
All seven files were resampled to 48 kHz stereo MP3. Tactical cues were silence-trimmed, equalized by role, and balanced to approximately -17.5 to -19 LUFS. The rain and fire recordings were reduced to stable excerpts, balanced near -25 LUFS, and rebuilt with a one-second end-to-start crossfade so repeated playback does not introduce a hard seam.
Battle playback uses track-specific loop trims for high-crest fire and mountain-wind recordings. Fallback volume is calculated from the measured integrated loudness of each active music-and-ambience pair, so quiet and loud BGM masters retain the same perceptual offset. Automated mix checks measure the actual encoded files across all 66 campaign battles and keep ambience 14-18 LU below its BGM; fire and snow-wind peaks must remain at least 1.5 dB below the paired music peak.
## Narrative Interface Cues
The source file in this section was downloaded from its official Pixabay detail page on 2026-07-22 and uses the Pixabay Content License linked above.
@@ -82,7 +84,7 @@ The following short movement clips were derived in-project on 2026-07-23 from th
| `src/assets/audio/sfx/hoof-wood-{1,2}.mp3` | Mounted movement through forest or across a wooden bridge | `src/assets/audio/sfx/horse-gallop.mp3` | DRAGON-STUDIO | https://pixabay.com/sound-effects/nature-horse-gallop-sfx-339736/ |
| `src/assets/audio/sfx/hoof-wet-{1,2}.mp3` | Mounted movement through river, water, marsh, or swamp | `src/assets/audio/sfx/horse-gallop.mp3` | DRAGON-STUDIO | https://pixabay.com/sound-effects/nature-horse-gallop-sfx-339736/ |
The derivatives are 0.34-0.48 second one-shots. They were excerpted from separate points in the parent recordings, resampled to 48 kHz stereo MP3, given 8 ms attack and 55 ms release fades, terrain-specific EQ, and conservative loudness/peak balancing. Runtime playback uses two-clip no-repeat pools per terrain and separate foot/hoof pools. Existing combat pools also use per-track gain compensation derived from integrated loudness measurements so randomized variants do not jump sharply in perceived level.
The derivatives are 0.34-0.48 second one-shots. They were excerpted from separate points in the parent recordings, resampled to 48 kHz stereo MP3, given 8 ms attack and 55 ms release fades, terrain-specific EQ, and conservative loudness/peak balancing. Runtime playback samples surfaces across the full movement route, treats rain and passable river-bank/ford tiles as wet, prevents the same active clip from doubling, and caps foot and hoof groups at two voices each. Existing combat pools also use per-track gain compensation derived from integrated loudness measurements so randomized variants do not jump sharply in perceived level.
## Sound Effects