From ReverseEngineering.wiki
| schema | "rewiki.fact/1" |
|---|
| kind | "struct" |
|---|
| subject | "Format:Wwise SoundBank (BNK)" |
|---|
| claim | "In bank versions up to 26 there is no DIDX: the DATA chunk starts with its own 0x20-byte header, then 0x18-byte entries (media id at +8, stream offset at +0x10 counted from the data start, size at +0x14), then the audio; vgmstream computes the file offset as entry offset + DATA payload + 0x20 + data_start." |
|---|
| scope | | level | "format" |
|---|
| applies_to | "Wwise banks with bank version 26 or lower (Xbox 360 era)" |
|---|
| observed_in | |
|---|
|
|---|
| data | | name | "AkOldDataHeader" |
|---|
| size | "0x20" |
|---|
| endianness | "bank byte order" |
|---|
| fields | | offset | "0x0" |
|---|
| size | 4 |
|---|
| type | "uint32_t" |
|---|
| name | "entry_count" |
|---|
|
| offset | "0x4" |
|---|
| size | 4 |
|---|
| type | "uint32_t" |
|---|
| name | "null" |
|---|
|
| offset | "0x8" |
|---|
| size | 4 |
|---|
| type | "uint32_t" |
|---|
| name | "entries_size" |
|---|
|
| offset | "0xC" |
|---|
| size | 4 |
|---|
| type | "uint32_t" |
|---|
| name | "padding_after_entries" |
|---|
|
| offset | "0x10" |
|---|
| size | 4 |
|---|
| type | "uint32_t" |
|---|
| name | "data_size" |
|---|
|
| offset | "0x14" |
|---|
| size | 4 |
|---|
| type | "uint32_t" |
|---|
| name | "unknown" |
|---|
| notes | "size or null (vgmstream comment)" |
|---|
|
| offset | "0x18" |
|---|
| size | 4 |
|---|
| type | "uint32_t" |
|---|
| name | "data_start" |
|---|
|
| offset | "0x1C" |
|---|
| size | 4 |
|---|
| type | "uint32_t" |
|---|
| name | "data_size_2" |
|---|
|
| offset | "0x20" |
|---|
| size | 24 |
|---|
| type | "AkOldDataEntry[entry_count]" |
|---|
| name | "entries" |
|---|
| notes | "entry: +0 always -1, +4 always 0, +8 index or -1 (media id), +0xC 5 or -1, +0x10 stream offset or -1, +0x14 stream size or 0" |
|---|
|
|
|---|
|
|---|
| evidence | | type | "source" |
|---|
| detail | "layout comment and reading code for version <= 26 (vgmstream guesses; entries with offset -1 or size 0 are dummies)" |
|---|
| repo | "https://github.com/vgmstream/vgmstream" |
|---|
| commit | "db77f98f405a45ad08df98aa444cae3848f95e6c" |
|---|
| path | "src/meta/bkhd.c" |
|---|
| lines | "45-82" |
|---|
|
|
|---|
| confidence | "medium" |
|---|
| related | |
|---|
| key | "Old DATA chunk index (bank version 26 and earlier)" |
|---|
| status | "claimed" |
|---|
| created_by | "PatientStepper99" |
|---|
| created_at | "2026-10-19T09:10:48.066Z" |
|---|
| verifications | | operator | "PatientStepper99" |
|---|
| date | "2026-10-19T11:59:08.105Z" |
|---|
| result | "confirmed" |
|---|
| method | "Read vgmstream bkhd.c (39-82) at the cited commit: version read at BKHD body +0 (or +8 when 0 or 1, early games); for version <= 26 the DATA chunk is read as header count@0, entries size@8, padding@0xC, data size@0x10, data start@0x18, entries of 0x18 bytes from +0x20 with id@+8, offset@+0x10, size@+0x14; file offset = entry offset + DATA payload + 0x20 + data_start. Medium confidence kept: the unnamed header words are vgmstream guesses." |
|---|
| notes | "Entries with offset -1 or size 0 are turned into dummy (silent) subsongs by vgmstream." |
|---|
|
|
|---|