From CloudModding OoT Wiki
(Redirected from Audiobank.zres)

Audiobank probably contains data related to playing sound effects, while Audiotable possibly contains the raw sound effects themselves. This page contains some notes on the structure of the Audiobank file created when running Zelda Resource Extractor on the Debug ROM. Beginning with NTSC 1.2, an additional 0xF370 bytes of padding were added to the end of the Audiobank file.

Beginning of file

The first 32-bit value is an offset of the list of drum-index pointers, and the second 32-bit value is an offset of list of soundeffect indexes. After that there is a series of 32-bit offsets to instrument indexes. If a 32-bit instrument value is empty that instrument index will be empty.

Audiobank Pointer Table

code (File)
Audiobank Pointer Table
VersionOffsetVRomVRam
Debug13827000BCC27000BCC690801550D0801554F0
NTSC 1.01026A000B896A000B89AC08011374080113B60
NTSC 1.110286000B8986000B89C808011390080113D20
NTSC 1.210271000B8971000B89B3080113DF080114210
PAL 1.00FFE6000B88E6000B892808011154080111960
PAL 1.10FFEA000B88EA000B892C080111580801119A0
JP GC101DC000B87DC000B881E080112CA0801130C0
JP MQ101DA000B87DA000B881C080112C80801130A0
USA GC101DA000B87DA000B881C080112C80801130A0
USA MQ101D8000B87D8000B881A080112C6080113080
PAL GC0FF59000B8759000B879B080110490801108B0
PAL MQ0FF57000B8757000B879908011047080110890
JP Collection101DA000B87DA000B881C080112C80801130A0

The Audiobank pointer table begins with an 0x10 byte header:

/* 0x00 */ short NumberOfInstrumentSets = 0x26;
/* 0x04 */ int Audiobank_VRomStartAddress; //ram only
/* 0x08-0x0F is padded to 0 */

Following that is a record for each set:

xxxxxxxx yyyyyyyy zzzzzzzz nn aa bb cc

  • x = Offset (ROM) or VRom Pointer (RAM) to the start of the instrument set within the Audiobank file.
  • y = Size of the instrument set in bytes.
  • z = Index for a list at 0xBCCD90(debug), or 0xB8A1B0(1.0) which gives the base address to load samples from.
  • n = Number of instruments to be read. Can be less than total but not greater, and does not include percussion.
  • a, b, c = Flags of some sort, setting a to 0x40 enables percussion, 0 disables it and c is the number of sound effects.

Unknown_1 structure

size example value meaning
32 bits 00 00 7F FB

00 00 7F BC

possibly a bit-field with flags
32 bits 00 00 2A 30 offset to structure Unknown_2
64 bits 00 00 00 00 00 00 00 00 Unknown, possibly padding
32 bits 00 00 01 80 offset to structure Unknown_3
32 bits 3F 30 66 66 Unknown
64 bits 00 00 00 00 00 00 00 00 Unknown, possibly padding

Unknown_2

Size: 16 bytes (128 bits) Example value: 00 01 7F BC 00 01 7F BC 7F BC 7F BC FF FF 00 00

Unknown_3

Size: 16 bytes

size example value meaning
32 bits 00 00 0A 8C Unknown
32 bits 00 03 EC E0 Unknown
32 bits 00 00 02 40 Offset to an Unknown_4 structure, which is usually (but not always) immediately after this Unknown_3.
32 bits 00 00 01 90 Offset to an Unknown_5 structure.

Unknown_4

size example value meaning
32 bits 00 00 00 00 Unknown
32 bits 00 00 12 C0 Unknown. Not an offset.
64 bits 00 00 00 00 00 00 00 00 Unknown

00 00 00 00 00 00 12 C0 00 00 00 00 00 00 00 00

Unknown_5 structure

size example value meaning
32 bits 00 00 00 02 a header of some sort, or maybe flags?
32 bits 00 00 00 02

00 00 00 04

block size, adjusts how long the data block is
X bits F8 72 F6 D0 FB F5 03 C3 08 64 06 A8 00 2C F9 EC

09 BB 04 48 FC 05 F7 1D F8 F4 FF D2 06 6F 08 00 F8 C0 F6 34 F9 53 FF DB 05 DB 08 0B 05 91 00 3D 0A D0 07 5E 00 29 F9 8A F7 20 F9 DB FF BD 05 37

size is 32x the block size defined by the previous set
64 bits 00 00 00 00 00 00 00 00 Very likely to be padding.

Channel Head

// -*- mode: c;-*-
typedef struct{
    local ubyte op;
    local int breakout = 0;
    local string warn;
    while ( breakout == 0 ){
        op = ReadUByte(FTell());
        switch( op ){
            case 0xC1:
                SHCommand C1(1);
                // C1 xx
                // Sets the instrument number, xx, to be used for the current channel.
                break;
            case 0xC2:
                SHCommand C2(1);
                // C2 xx
                // Transposition, signed. xx is the number of semitones to transpose by.
                break;
            case 0xC4:
                SHCommand C4(0);
                // Initialises the channel for music playback.
                break;
            case 0xD3:
                SHCommand D3(2);
                // D3 xx (yy)
                // Pitch bend amount, signed, value of xx. yy is a timestamp used between control changes.
                break;
            case 0xD4:
                SHCommand D4(1);
                // D4 xx
                // Effects level (echo). xx is the effect amount.
                break;
            case 0xD8:
                SHCommand D8(2);
                // D8 xx (yy)
                // Vibrato amount, value of xx. Higher values can produce odd sounds. yy is a timestamp used between control changes.
                break;
            case 0xDD:
                SHCommand DD(2);
                // DD xx
                // Channel pan. xx is the pan amount, 00 = hard left, 3F = centre, 7F = hard right.
                break;
            case 0xDF:
                SHCommand DF(2);
                // xx (yy)
                // Channel volume. xx is the volume value, yy is a timestamp used between control changes (when the volume is changing constantly).
                break;
            case 0xE9:
                SHCommand E9(1);
                // E9 xx
                // Priority. Unknown how it is used, takes the value of xx.
            case 0x80: // unused?
            case 0x81: // unused?
            case 0x82: // unused?
            case 0x83: // unused?
            case 0x84: // unused?
            case 0x85: // unused?
            case 0x86: // unused?
            case 0x87: // unused?
            case 0x88:
            case 0x89:
            case 0x8A:
            case 0x8B:
            case 0x8C: // unused?
            case 0x8D: // unused?
            case 0x8E: // unused?
            case 0x8F: // unused?
                SHCommand EIGHTX(2);
                // 8x yyyy
                // Points to music data to be played on the current channel. x is the 'note layer' to be used (8 - B), up to a maximum of four note layers can be loaded per channel. yyyy is the offset of the music data to be played, relative to the start of the sequence file.
                break;
            case 0xFB:
                SHCommand FB(2);
                // xxxx
                // Offset to loop from, xxxx is the offset relative to the start of the sequence file.
                break;
            case 0xFD:
                // *** slave copy from SeqHead ***
                if( ReadUByte(FTell()+1) > 0x7F ){
                    SHCommand FD(2);
                } else {
                    SHCommand FD(1);
                }
                // FD xx / FD yyyy
                // Timestamp (number of 'ticks' to wait before the next command is read, relative to tempo), variable length. If xx goes above 7F, add 8000 to it to get yyyy.
                break;
            case 0xFF:
                // FF
                // Marks the end of the channel header. 
                SHCommand FF(0);
                //breakout = 1;
                break;
            default:
                SHCommand IDK(1);
                SPrintf(warn, "\nWARNING: Unknown ChanHead opcode (0x%02L)!", op);
                EXWarning(warn);
                break;
        }
    }
} ChanHead;
(Source: DerrikeG)


Instrument

// -*- mode: c;-*-
/*
    COVERAGE (ITEMS DONE):
    ITEMS NOT DONE:
    TableSample
    Unknown2
    SampleBlock
    SampleHead
    InstrumentSample
    ElusiveRecord2
    ElusiveRecord1
    InstProperties
    PercussionInstrumentIndex;
    PercussionInstrument;
    OtherInstrument
*/

typedef struct(ubyte sampleBank){
    SIBE;
    SIT;
    ubyte unknown1 : 8;
    if( unknown1 != 0 ){
        Printf("WARNING: The first byte of a TableSample wasn't 0 (was %d) @ 0x%Lx\n", unknown1, FTell()-8);
    }
    SIT;
    uint sampleSize : 24;
    SIT;
    uint offset; // from the Audiotable
    SAVE;
    FSeek( AUDIOTABLE + sampleTable.mySampRecords[sampleBank].offset + offset );
    SIT;
    ubyte audioSample;
    //ubyte audioSample[sampleSize];
    REST;
    SIEND;
} TableSample;

// deathbasket structure 1
typedef struct(uint seekPos){
    SIBE;
    SIT;
    uint unknown1;
    SIT;
    uint unknown2;
    if( unknown1 == 0 ){
        SIT;
        uint zeroes[2];
    } else {
        SIT;
        uint FFs;
        SIT;
        uint zeroes;
        SIT;
        uint unknown3[8];
    }
    SIEND;
} Unknown2;

// deathbasket structure 2
typedef struct(uint seekPos){
    SIBE;
    SIT;
    uint type;
    SIT;
    uint blockSize;
    SIT;
    ubyte dataBlock[32 * blockSize];
    SIEND;
} SampleBlock;

typedef struct(uint seekPos, ubyte sampleBank){
    SIBE;
    TableSample sampleFromTable(sampleBank);
    SIT;
    uint unknownPointer; //chunkFooter?!
    SIT;
    uint chunkPointer;
    SAVE;
    FSeek( seekPos + chunkPointer);
    SampleBlock sampleData(seekPos);
    REST;

    SAVE;
    FSeek( seekPos + unknownPointer );
    Unknown2 unknown2(seekPos);
    REST;
    SIEND;
} SampleHead;

typedef struct(uint seekPos, ubyte sampleBank){
    SIBE;
    SIT;
    uint sampleHead;
    SIT;
    float pitch;
    SIT;
    if( sampleHead > 0 ){
        SAVE;
        FSeek( seekPos + sampleHead );
        SampleHead headSample(seekPos, sampleBank);
        REST;
    }
    SIEND;
} InstrumentSample;

// elusive record 2
typedef struct(uint seekPos, ubyte sampleBank){
    SIBE;
    SIT;
    ubyte unknown1 : 8;
    SIT;
    uint  unknown2 : 24;
    InstrumentSample sample(seekPos, sampleBank);
    SIT;
    uint  unknown5; // if it's a set-local pointer,
                    // it's an instrumentProperties pointer
    SIEND;
} ElusiveRecord2;

// elusive record 1
typedef struct(uint seekPos, ubyte sampleBank){
    SIBE;
    local uint64 sizeToRead = (parentof(this).seek2 - parentof(this).seek1) / 4;
    SIT;
    uint er2Pointers[sizeToRead];
    //SIT; // flip it so we don't odd-man-out the next var
    local uint64 i;
    for( i = 0; i < sizeToRead; i++ ){
        if( er2Pointers[i] > 0){
            SAVE;
            FSeek( seekPos + er2Pointers[i] );
            ElusiveRecord2 er2(seekPos, sampleBank);
            REST;
        }
    }
    SIEND;
} ElusiveRecord1;

typedef struct{
    SIBE;
    SIT;
    ushort props[8];
    SIEND;
} InstProperties;

// everything below this definitely belongs in this file

typedef struct(uint seekPos, ubyte sampleBank){
    SIBE;
    SIT;
    ubyte unknown1[4];
    InstrumentSample sample(seekPos, sampleBank);
    SIT;
    uint instPropertiesPointer;
    /*
        xxxxxxxx zzzzzzzz pppppppp yyyyyyyy
         
        x = Unknown.
        zzzzzzzz = Pointer to unknown data structure.
        pppppppp = Default sample pitch.
        yyyyyyyy = Pointer to instrument properties.
    */
    /*
    if( pointerToUnknown > 0 ){
        SAVE;
        FSeek( seekPos + pointerToUnknown );
        SampleHead headSample(seekPos);
        REST;
    }
    */
    SAVE;
    FSeek( seekPos + instPropertiesPointer );
    InstProperties properties;
    REST;
    SIEND;
} PercussionInstrumentIndex;

typedef struct(uint seekPos, ubyte sampleBank){
    SIBE;
    SIT;
    local uint64 numRecords = 64;
    local uint64 j;
    uint instrumentRecords[numRecords];
    for( j = 0; j < numRecords; j++ ){
        SAVE;
        FSeek( seekPos + instrumentRecords[j] );
        PercussionInstrumentIndex myInstruments(seekPos, sampleBank);
        REST;
    }
    SIEND;
} PercussionInstrument;

typedef struct(uint seekPos, ubyte sampleBank){
    SIBE;
    SIT;
    ubyte startNoteNumbers[3];
    SIT;
    ubyte maybeDecay;
    SIT;
    uint instPropertiesPointer;
    SAVE;
    FSeek( seekPos + instPropertiesPointer );
    InstProperties properties;
    REST;
    InstrumentSample samples(seekPos, sampleBank)[3]<optimize=false>;
    SIEND;
} OtherInstrument;
(Source: DerrikeG)


Instrument Set

// -*- mode: c;-*-
/*
    COVERAGE (ITEMS DONE):
    InstrumentSetTable
    InstrumentSet
    ITEMS NOT DONE:
    InstrumentSetData
*/

typedef struct {
    SIBE;
    local uint64 i;
    /* 
    the value of Seek1 is valid relative to itself,
    wherein it points to a list of relative pointers
    */
    SIT;
    uint seek1;

    SIT;
    uint seek2;

    if(parentof(this).unknown1.bit2 || parentof(this).unknown1.bit4 || parentof(this).unknown1.bit6 || parentof(this).unknown1.bit8){
        AssertUIntGTZero(seek1, "InstrumentSetData->seek1");

        SAVE;
        FSeek( startof(this) + seek1 );
        if(!parentof(this).unknown1.bit2){
            ElusiveRecord1 er1(startof(this), parentof(this).sampleBank);
        } else {
            PercussionInstrument myInstruments(startof(this), parentof(this).sampleBank);
        }
        REST;
    }


    if(parentof(this).unknown1.bit4 || parentof(this).unknown1.bit6 || parentof(this).unknown1.bit8){
        AssertUIntGTZero(seek2, "InstrumentSetData->seek2 [non-drums do not blank]");
        SAVE;
        FSeek( startof(this) + seek2 );
        for( i = 0; i < numExtraInstrumentSamples; i++ ){
            InstrumentSample extraInstrumentSamples(startof(this), parentof(this).sampleBank);
        }
        REST;
    } else if(parentof(this).unknown1.bit2){
        AssertNumberEquals(seek2, 0, "InstrumentSetData->seek2 [drums should be blank]");
    }

    SIT;
    uint myInstrumentLocations[ parentof(this).numInstruments ];
    
    // iterate through all the instruments
    for( i = 0; i < parentof(this).numInstruments; i++ ){
        if( myInstrumentLocations[ i ] != 0 ){
            SAVE;
            FSeek( startof(this) + myInstrumentLocations[ i ] );
            //FSeek( AUDIOBANK + aBankOffset + myInstrumentLocations[ i ] );
            OtherInstrument myInstruments(startof(this), parentof(this).sampleBank);
            REST;
        } else {
            //AssertUIntGTZero( myInstrumentLocations[ i ], "InstrumentSetData->myInstrumentLocations[]" );
        }
    }
    SIEND;
} InstrumentSetData;

typedef struct {
    SIBE;
    SIT;
    uint offset; // relative to AUDIOBANK
    SIT;
    uint size;
    SIT;
    ubyte majorVersion;
    //AssertNumberEquals(majorVersion, 2, "InstrumentSet->majorVersion");
    SIT;
    ubyte minorVersion;
    AssertNumberInBetween(minorVersion, 0, 2, "InstrumentSet->minorVersion");
    SIT;
    ubyte sampleBank<comment="This corresponds with SampleTable->mySampRecords[#]">;
    AssertNumberInBetween(sampleBank, 0, sampleTable.numSamps, "InstrumentSet->sampleBank");
    SIT;
    ubyte unknown2;
    AssertNumberEquals(unknown2, 0xFF, "InstrumentSet->unknown2 (NOT PADDING!!!)");
    /*
    The third byte is an index for a list at 
    SAMPLES which gives the base address to load
    samples from. What is the purpose of this? Well
    apparently just to be a pain in the arse, 
    because taking these numbers and adding them to sample 
    pointers will load that sample perfectly if the index
    is changed to 1, which has a base address of zero and 
    is what most instrument sets use. Only four sets seem 
    to actually do this though, namely the:
    Deku Tree, Jabu-Jabu's belly, Goron City and the Spirit Temple.
    Super Mario 64 does something similar but it affects every single instrument set in the game.
    */
    //Number of instruments to be read (can be less than total but 
    // not greater, does not include percussion).
    SIT;
    ubyte numInstruments;
    AssertUIntGTZero(numInstruments, "InstrumentSet->numInstruments");
    SIT;
    BitEight unknown1;         // aa
    AssertNumberEquals(unknown1.bit1, 0, "InstrumentSet->unknown1->bit1");
    // AssertNumberEquals(unknown1.bit2, 0, "InstrumentSet->unknown1->bit2");
    AssertNumberEquals(unknown1.bit3, 0, "InstrumentSet->unknown1->bit3");
    //AssertNumberEquals(unknown1.bit4, 0, "InstrumentSet->unknown1->bit4");
    AssertNumberEquals(unknown1.bit5, 0, "InstrumentSet->unknown1->bit5");
    //AssertNumberEquals(unknown1.bit6, 0, "InstrumentSet->unknown1->bit6");
    AssertNumberEquals(unknown1.bit7, 0, "InstrumentSet->unknown1->bit7");
    //AssertNumberEquals(unknown1.bit8, 0, "InstrumentSet->unknown1->bit8");
    /*
    aa = a bitrange
    only bits 6 8 and 2 are used at any given time
    bit 2 handles percussion
        * dOoTMQ_E_debug
        * dMM_E_debug
    bit 4 is only ever set on the first record
        * dMM_E_debug
    bit 6 is only ever set on the first record
        * dOoTMQ_E_debug
    bit 8 is only ever set on the second record
        * dOoTMQ_E_debug
        * dMM_E_debug
    */

    SIT;
    ushort numExtraInstrumentSamples;

    /*
    foreach on the number of instruments
    */
    SAVE;
    FSeek( AUDIOBANK + offset );
    InstrumentSetData myData;
    REST;
    SIEND;
} InstrumentSet;

typedef struct {
    SIBE;
    SIT;
    ushort numInstrumentSets;
    SIT;
    ubyte  instrumentSetPadding[14];
    AssertUBytesNull( instrumentSetPadding, "instrumentSetPadding" );

    EXWarning("OPTIMIZE FIX: InstrumentSetTable->myInstrumentSets (InstrumentSet)");
    InstrumentSet myInstrumentSets[numInstrumentSets]<optimize=false>;
    SIEND;
} InstrumentSetTable;
(Source: DerrikeG)


Sample

// -*- mode: c;-*-
typedef struct {
    SIBE;
    SIT;
    uint offset; // relative to AUDIOTABLE
    SIT;
    uint size;
    SIT;
    ubyte unk1;
    AssertNumberEquals( unk1, 2, "SampRecord->unk1" );
    SIT;
    ubyte unk2;
    AssertNumberEquals( unk2, 4, "SampRecord->unk2" );
    SIT;
    ubyte zeroes[6];
    AssertUBytesNull( zeroes, "SampRecord->zeroes" ); 
    SIEND;
} SampRecord;

typedef struct {
    SIBE;
    SIT;
    ushort numSamps;
    SIT;
    ubyte  numSampsPadding[14];
    AssertUBytesNull( numSampsPadding, "numSampsPadding" );

    SIT;

    EXWarning("OPTIMIZE FIX: SampleTable->mySampRecords (SampRecord)");
    SampRecord mySampRecords[numSamps]<optimize=false>;
    SIEND;
} SampleTable;
(Source: DerrikeG)


Sequence

// -*- mode: c;-*-
struct SeqHead;

local uint doWalkThrough = 0;
local ushort seqNo = 0;
typedef struct {
    SIT;
    /* offset relative to start of audioseq file */
    uint relative_offset;     
    SIT;
    /* length in bytes */
    uint length;              
    SIT;
    /* 0x02000000, 0x02010000, 0x02020000 */
    uint seq_type;            
    SIT;
    ubyte padding[4];             
    AssertUBytesNull( padding, "SequenceRecord->padding" ); 

    SAVE;
    SetBackColor( cPurple );
    FSeek(AUDIOSEQ + relative_offset);
    SeqHead microcode;
    REST;
} SequenceRecord;

typedef struct {
    SIT;
    ushort numSeqs;
    SIT;
    ubyte  numSeqsPadding[14];
    AssertUBytesNull( numSeqsPadding, "numSeqsPadding" );

    EXWarning("OPTIMIZE FIX: SequenceTable->mySequenceRecords (SequenceRecord)");
    SequenceRecord mySequenceRecords[numSeqs]<optimize=false>;
} SequenceTable;
(Source: DerrikeG)


Sequence Head

// -*- mode: c;-*-
typedef struct (uint noArgs){
    SIT;
    ubyte command;
    local uint64 i;
    for( i = 0; i < noArgs; i++ ){
        SIT;
        ubyte arg;
    }
} SHCommand;

typedef struct {
    /*
        Sequence Header Commands[edit]
        
        Note: some sequences have the following data after the first four bytes of their sequence header:
        D7FFFF 8776CCFF7786F3 vv F2 ww C801F3 xx [C801FA yyyy] FB zzzz
        v, w and x are unknowns, y and z seem to be pointers to 9x commands for different parts of the sequence within the sequence header. The part in square brackets is repeated in some sequences. Its purpose is currently unknown but it may be to do with continuing playback mid-sequence after returning to the area from a house/shop.
        
        https://github.com/sauraen/seq64/blob/master/romdesc/mm_info.xml
        https://sites.google.com/site/messiaen64/toads-music-incomplete
        https://sites.google.com/site/messiaen64/differences-between-sseq-nds-and-mario64-seq-format
        https://sites.google.com/site/messiaen64/mario-64-sequenced-music-specification
    */
    local ubyte op;
    local ubyte breakout = 0;
    local string warn;
    while ( breakout == 0 ){
        op = ReadUByte(FTell());
        switch( op ){
            case 0xCC:
                SHCommand CC(2);
                // yyxx
                // Unknown. yy starts at zero and increases each time the command appears, xx is (always?) 73.
                break;
            case 0xD3:
                SHCommand D3(1);
                // xx
                //Seems to be something to do with the sequence type or format; for Super Mario 64, xx is usually 80, for Zelda 64, xx is usually 20.
                break;
            case 0xD5:
                SHCommand D5(1);
                // xx
                // Unknown. xx usually takes the value of 32. For sequences that don't play any music xx is 46.
                break;
            case 0xD6:
                SHCommand D6(2);
                // xxxx
                // Disables channels given by xxxx.
                break;
            case 0xD7:
                SHCommand D7(2);
                // xxxx
                // Enables channels given by xxxx (boolean, each channel is one bit).
                break;
            case 0xDB:
                SHCommand DB(1);
                // xx
                // Master volume control, xx is the volume.
                break;
            case 0xDD:
                SHCommand DD(1);
                // xx
                // Tempo control, xx is the tempo value in beats per minute.
                break;
            case 0x90:
            case 0x91:
            case 0x92:
            case 0x93:
            case 0x94:
            case 0x95:
            case 0x96:
            case 0x97:
            case 0x98:
            case 0x99:
            case 0x9A:
            case 0x9B:
            case 0x9C:
            case 0x9D:
            case 0x9E:
            case 0x9F:
                SHCommand NINEX(2);
                // yyyy
                // Points to a channel header offset. 9X is the channel number (0 - F) and yyyy is the offset of that channel's header relative to the start of the sequence file.
                break;
            case 0xFB:
                SHCommand FB(2);
                // xxxx
                // Offset to loop from, xxxx is the offset relative to the start of the sequence file.
                break;
            case 0xFD:
                if( ReadUByte(FTell()+1) > 0x7F ){
                    SHCommand FD(2);
                } else {
                    SHCommand FD(1);
                }
                // FD xx / FD yyyy
                // Timestamp (number of 'ticks' to wait before the next command is read, relative to tempo), variable length. If xx goes above 7F, add 8000 to it to get yyyy.
                break;
            case 0xFF:
                SHCommand FF(0);
                breakout = 1;
                break;
            default:
                SHCommand IDK(1);
                SPrintf(warn, "WARNING: Unknown SeqHead opcode (0x%02LX)!", op);
                EXWarning(warn);
                break;
        }
    }
} SeqHead;
(Source: DerrikeG)