gora/UnheardCity
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Version 3.2.1: whole bank copies of voice lines, unique names, docs figures checked by tools/check_docs.py

Browse Source
This commit is contained in:
Cristian Gora
2026-10-10 19:30:06 +02:00
parent 1586abb8e9
commit cac52e7597
29 changed files with 919 additions and 188 deletions
Download Patch File Download Diff File Expand all files Collapse all files

58
docs/formats/banks.md
View File

@@ -10,29 +10,31 @@ Survey of the installed game (buildid 3718300), 2026-10-09.
| where | banks | unique in-memory WEMs |
|-------|-------|-----------------------|
| loose `SFX\*.bnk` (Batman/Catwoman footsteps, 32 surfaces each, + `TST_Prep_Media`) | 65 | 1,117 |
| banks embedded in `.upk` packages (in 1,432 of the 2,650 packages) | 3,340 | 15,641 (~276 MB) |
| banks embedded in `.upk` packages (in 1,432 of the 2,650 packages) | 3,340 | 14,802 |
All of them are in the catalog (archive `BANK`).
The two sets share 119 media IDs: 15,800 unique media IDs in 96,011 places (bank copies). 396 of them are
**prefetch heads** of streamed WAD sounds (below), so there are **15,404 in-memory sounds**. All copies of a
**prefetch heads** of streamed WAD sounds and 49 are **whole copies** of WAD recordings (below), so there are
**15,355 in-memory sounds**. All copies of a
media ID are byte-identical except `1C898731` (two variants). The WADs, by comparison, are fully covered: the
BKPK tables in the packages reference 2,543 WAD entries, all of them in the catalog.
Embedded WEMs by bank name prefix: `Speech` 7,872, `CHR` (character foley, combat, bosses) 3,896, `PSO` 781,
`WPN` 664, `NPC` 475, `OBJ` 439, `SEQ` 265, `GUI` 208, `PSA` 203, `MTN` 196, `PX` 195, `DVC` 100, `GEN` 80,
`PHY` 73, `AMB` 65, `SBJ` 50, others < 30. The `Speech_` banks (e.g. `Speech_EMOTE_Batman`, `..._GRUNT`) hold
speech: grunts and efforts (language-neutral, 165) and **localized lines** (about 1,460 per language, see
*Localized banks*).
All WEMs seen are Vorbis (format tag 0xFFFF), the same codec as the WADs: the footsteps are mono (4 stereo),
the embedded ones mono or stereo (~22 % stereo in a 2,533-WEM sample).
Embedded WEMs by bank name prefix (a media held by banks of two prefixes counts for both): `Speech` 7,827, `CHR`
(character foley, combat, bosses) 3,624, `PSO` 737, `WPN` 577, `NPC` 471, `OBJ` 411, `SEQ` 253, `PSA` 203,
`PX` 195, `MTN` 191, `GUI` 155, `DVC` 100, `GEN` 80, `PHY` 73, `AMB` 61, `SBJ` 50, others < 30. The `Speech_` banks (e.g. `Speech_EMOTE_Batman`, `..._GRUNT`) hold
speech: efforts, grunts and a few other lines stored once for every language (119 sounds) and **localized
lines** (about 1,450 per language, see *Localized banks*).
All WEMs seen are Vorbis (format tag 0xFFFF), the same codec as the WADs: the loose footsteps are mono (4
stereo); of all 15,355 in-memory sounds 13,982 are mono, 1,196 stereo, 177 have 3-6 channels.
## Names
- No list names these media IDs (the community hash list names only WAD entries and banks).
- Media IDs have no names of their own: a sound is named after the Wwise event that plays it (event names
are in the packages' name and export tables, event ID = FNV-1 of the name), else after its bank.
- `CookedPCConsole\CookedBanks.txt` lists all 3,345 cooked bank names. Bank ID = FNV-1 32 of the lowercased
name (the same hash Wwise uses for every name). 3,338 of them are embedded in packages, 4 are loose `.bnk`
files (`Init`, `Global`, `Music`, `Music_PDLC`); `AMB_OW_A6`, `AMB_OW_C1`, `AMB_OW_E1`,
`Speech_EMOTE_Thug_Muffled_Stop` and `Speech_GFxUI_ArkhamCityStories` were found nowhere.
name (the same hash Wwise uses for every name). 3,338 of them are embedded in packages (`Music`, `Music_PDLC`
also as loose `.bnk` files), 2 are only loose `.bnk` files (`Global`, `Init`); `AMB_OW_A6`, `AMB_OW_C1`,
`AMB_OW_E1`, `Speech_EMOTE_Thug_Muffled_Stop` and `Speech_GFxUI_ArkhamCityStories` were found nowhere.
- Loose footstep banks are named in their STID chunk (`CHR_BatmanFootsteps_Concrete_Media`, ...).
## Bank layout (version 56, Wwise 2011.1)
@@ -51,6 +53,12 @@ with bank id 0 whose media is not in the same bank (loaded by another bank); 133
(`wad.md`). A prefetch source's media is also in the bank's DIDX: that in-memory copy is the **head** of the
streamed WEM (its size = the source's size), which the game plays while the rest is read from the WAD. 396
WAD sounds have such heads (2,101 copies); replacing one of them must rewrite the heads too.
49 more WAD recordings (9-11 per language: game-over taunts, an encryption-card line, a cutscene line, scripted
story lines) are
also stored **whole** in banks: an in-memory source with the WAD recording's own media ID, byte-identical to
the WAD WEM (84 copies). They are handled like heads: part of the WAD line, written and reverted with it, not
sounds of their own (`catalog._whole_recordings`). Heads and whole copies together: 445 recordings, 2,185
copies (`BANK_locations.csv.gz` kind `p`).
## Localized banks
Speech banks are localized inside the (language-neutral) packages: one `AkBank` export starts with a language
@@ -58,9 +66,10 @@ list (FStrings `INT FRA ITA DEU ESN`) and then holds one block per language, in
of the speech WAD entry for that language (`wad.md`) and a complete bank, with the same bank ID in every
language. ackit gives each bank and each BKPK table the language of its position in that list (an export
whose block count differs from its language count would be reported; none does). Each in-memory media gets
the languages of the banks that hold it (`BANK.csv` column `language`): 8,098 sounds are language-neutral,
~1,460 belong to one voice language, 6 to several (a DLC intro line shared by INT/FRA/ITA/ESN while DEU has
its own); a media stored in the banks of every language is listed as `''` (the same recording everywhere).
the languages of the banks that hold it (`BANK.csv` column `language`): 8,092 sounds are language-neutral,
1,450-1,456 belong to one voice language, 6 to several (5 shared by INT/FRA/ITA/ESN while DEU has its own, 1
by INT/FRA/ITA); a media stored in the banks of every language is listed as `''` (the same recording
everywhere; 167 of these are voice lines, e.g. efforts and radio chatter).
A catalog shows the neutral sounds plus the loaded language's lines; names are unique per language.
## Lines: one recording per language
@@ -73,17 +82,18 @@ and the game ties the recordings together itself:
- Every recording, streamed, prefetched or in memory, is the source of a Wwise Sound object; Sound objects,
events, actions and containers have the same IDs in every language, only the media differ. A *line* = the
recordings that are sources of the same Sound objects: 12,434 lines (WAD and in-memory), each with exactly
one recording in each of the 5 voice languages (61,510 recordings), no Sound object with two recordings in
one recording in each of the 5 voice languages (61,485 recordings), no Sound object with two recordings in
one language. Line identity (`line` column) = the smallest Sound object ID.
- Prefetched sources count too: the 396 lines with a prefetch head (mostly the game-over taunts) are linked
only through them.
- Prefetched sources count too: the 81 lines with a prefetch head (396 recordings over the 5 languages; mostly
the game-over taunts) are linked only through them. A WAD recording that a bank also stores whole (49) is
the same recording (same media ID), so its line is unchanged and the bank copies are written with it.
- 5 recordings per language are stored twice in that language's WAD (same media ID in two entries, byte-
identical, e.g. `Speech_STEEL_A1_CH2_Thug_behinddoormuffled` and `DIA_SteelA1Muffled`): the second place
is a copy of the line (same `line`, keeps its position name) and is written and reverted with it, because
the game plays whichever place the loaded package's table names.
- Independent checks (no BKPK involved): every WAD recording is named as a source by a Sound object of a bank
of its own language (54,055 of 54,055); every stored prefetch head equals the start of the recording its
line gives it (2,101 of 2,101); a line's recordings have similar lengths across languages (11,761 of 12,434
of its own language (54,055 of 54,055); every stored head or whole copy equals the start of the recording
its line gives it (2,185 of 2,185); a line's recordings have similar lengths across languages (11,761 of 12,434
within a factor 1.5; the others are different edits, e.g. a 0.55 s ESN line vs 5.46 s elsewhere).
## Banks inside packages
@@ -99,10 +109,12 @@ and the game ties the recordings together itself:
## How ackit edits them
- Catalog archive `BANK` (one row per media ID) + `data/catalog/BANK_locations.csv.gz` (every copy, and the
prefetch heads of speech lines); built by `tools/build_catalog.py`, checked by `ackit verify --catalog`.
heads and whole copies of WAD speech recordings); built by `tools/build_catalog.py`, checked by `ackit verify
--catalog`.
- A replacement is written **in place into every copy** (no larger than the smallest copy, zero-padded), so
DIDX/HIRC offsets and sizes and the package layout never change. Loose banks are patched directly; in
packages only the touched blocks are re-encoded (`upk.md`). A sound left in one package only would let the
game load the original from another one.
- Replacing a speech line that has prefetch heads rewrites the heads (first bytes of the new, zero-padded line).
- Replacing a speech line that has prefetch heads or whole copies rewrites them (the first bytes of the new,
zero-padded line; for a whole copy, all of it).
- Design and revert rules: `docs/decisions.md` (4b). The same WEM rules apply (`wem.md`).