.fat sound banks

A sound bank packs RIFF/WAVE files into one file. resource.hog holds most of them; CD1.HOG and CD2.HOG each add VRSFX.FAT and VRSND.FAT, and a copy of wlksmp.fat identical to resource.hog's.

sltool fat ls <bank>                  # every sound: offset, size, priority, format, length, and the
                                      # 3D sounds that play it, for smp3d.fat
sltool fat extract <bank> <out-dir>   # every sound as a WAV file, such as stdsmp_012.wav
make sounds                           # every sound in resource.hog into game/sounds

Layout

Offset Size Field
0 4 2.00, four raw bytes
4 4 Sound count
8 12 x count Entries
Offset Field
0 Offset of the sound from the start of the bank
4 Size
8 Priority

The sounds follow the entries back to back, to the end of the file, and each is a complete WAVE file (formats/wave.zig) whose RIFF length matches its entry's size. All but one are IMA ADPCM (format 0x11, 4 bits per sample, with a fact chunk giving the length) at 11,025, 22,050 or 44,100 Hz, mono or stereo; the other is 16-bit PCM. sltool fat extract writes them unchanged.

Version 1.00

Privateer 2, StarLancer's predecessor, keeps its sounds in banks of the format's first version (Privateer 2). sltool fat ls lists them, and sltool fat extract writes their sounds as stored. The header is the same, with 1.00 as its magic. Each entry is 16 bytes:

Offset Size Field
0 4 Offset of the sound from the start of the bank
4 3 The sound's size decoded, in bytes
7 1 Unknown: 1 in every shipped bank
8 4 Priority, 10 in every shipped bank
12 4 Unknown: 01 00 01 00 in the speech banks, and 01 00 01 03 in the others

An entry gives no stored size, so a sound runs to the next sound or to the end of the bank. Every shipped bank holds one sound. The sounds aren't WAVE files but raw 4-bit samples: each decodes to four times its stored size in 16-bit samples. Unknown: the codec. Decoded as IMA, OKI or Yamaha ADPCM, the samples drift far from zero, which recordings don't (#1074).

Playback

The engine reads a whole bank into memory with its generic file loader, hog_read_file (0x004C7F60), and plays sound n of it with sound_play (0x00481F80), which passes Miles the WAVE file at bank + offset. The priority decides whether a sound may take over a voice already playing: the player finds the busy voice of lowest priority and stops it only for a sound whose priority is higher, then records the new sound's priority on the voice (Sound).

The shipped banks use priorities 1, 5, 50 and 10000. Most hold a single sound: those named for the flyable fighters, such as PREDATOR.FAT, each hold one at priority 10000.

The banks

The payload loads these banks by name:

Bank What it holds
stdsmp.fat The sounds of the menus and the flight display, which scripts can also play by number, and the sounds of a frame placed around the camera (Sound)
smp3d.fat The 3D sounds: guns, missiles, explosions, engines, jumps and more, which sltool fat ls names (3D sounds)
betty.fat Betty's warnings in the cockpit
ldsmp.fat The loadout's sounds
inter\itac\itacsnd.fat The ITAC's sounds
vrsnd.fat, wlksmp.fat A carrier's rooms: the ship's hum and the ways into some views; the player's steps and the doors
vrsfx.fat, waitloop.fat The briefing: the doors on the way in and the room's chatter; the sound that plays while it loads
newsloop.fat, rlanda.fat to rlande.fat, ylanda.fat to ylande.fat The landings: the news reports' loop, and a sound for each outcome of a landing on the Reliant or the Yamato

Mods' sounds

A mod's sound replaces one of a bank's sounds. It's named after the bank and the sound's number, as sltool fat extract saves it (fat.soundName), and is a WAVE, FLAC, MP3, Ogg Vorbis or Opus file (Sounds). When OpenReliant reads a bank, it builds a new one in memory with the mods' sounds instead of the bank's own, each entry keeping its priority (fat.rebuild, hog_snd.BankFile.of), so the game plays them like its own. If the drivers can't play a mod's sound, the bank's own sound stays, and the log says so.

Improvement: the original plays only its banks' own sounds.

Edit this page on GitHub. The documentation is under CC BY-SA 4.0.