Sound Tracker Module File Format
--------------------------------
This document (C) Copyright 1990, Steven W Green. This document may
be freely distibuted, providing that any corrections are sent to me,
this copyright notice remains intact and that it does not breach any
law.
Deduced by reverse engineering a soundtracker player and looking at
the files. The author of this document can not be held responsible
for any damage caused by inacuracies in this document. The author has
absolutely no relationship with anybody concerned with the creation of
SoundTracker or related utilities.
Overall Structure:
Module name char[20];
Instruments INSTRUMENT[15 or 31] 30 bytes per instrument
Header UBYTE[2];
Sequences UBYTE[80];
[ID] LONG 'M.K.'
Patterns PATTERN[];
Samples WORD[][]
There appear to be 2 different flavours of Soundtracker modules.
Earlier ones, which allow 15 instruments and later ones allow 31. A
way to find out the version is to look at the LongWord at offset $438
from the start of the file and compare it with the packed string
'M.K.' (presumably the initials of someone?).
Version1 Version2
offset length | offset length | section
--------------+---------------+-----------------
0 20 | 0 20 | Module Name
$14 30X15 | $14 30X31 | Instruments
$1d6 2 | $3b6 2 | Header
$1d8 80 | $3b8 80 | Sequences
| $438 4 | Version ID 'M.K.'
$258 1024X? | $43c 1024X? | Patterns
? ? | ? ? | Samples
--------------+---------------+-----------------
Instrument structure
--------------------
Each instrument occupies 30 bytes. There are either 15 or 31
instruments in a module.
char name[20]; A null terminated filename
WORD unknown;
WORD length; Length of sample in words
WORD volume; Default volume (0..64)
WORD unknown
WORD repeat; 1=No repeat, otherwise position in sample for
repeat play of continuous sample.
To find the sample, you must find the end of the track data by finding
the maximum track number from the seqence section and then multiplying
by 1024 and adding to the start of the pattern data. Then each
instrument sample is simply stored one after the other.
Note that the 1st instrument is instrument 1. Thus the instruments
are numbered 1..15 or 1..31.
The instruments should be sampled as an A (110/220/440Hz) at 14KHz.
This is a calculated guess.
The filename is an Amiga name, often beginning with a volume name such
as ST-01:filename
Header
------
UBYTE n_tracks; Number of tracks to play
UBYTE unknown; Maybe highest note (78)
Sequences
---------
This is an array of 80 bytes. It is a list of patterns to play and
they are read in order for n_tracks tracks, unless a jump instruction
is encountered in the pattern data.
Pattern Data
------------
Each Pattern is 1024 bytes long. A pattern consists of 64 events.
there is one Longword for each of the 4 tracks. An event is executed
each 'tempo' vertical-blanks. I don't know whether this should be 50Hz
or 60Hz. I shall be assuming 50Hz. The default tempo is 6. A
typical song may have 4 events per crotchet => 16 per bar of 4/4 => 4
bars per pattern. Then on 60Hz systems you could skip one in 6 VBL's
to keep the same tempo.
The format of a track event is:
7 6 5 4 3 2 1 0
+---+---+---+---+---+---+---+---+
Byte 0 | Unknown | I | Frequency hi |
+---+---+---+---+---+---+---+---+
+---+---+---+---+---+---+---+---+
Byte 1 | Frequency mid | Frequency lo |
+---+---+---+---+---+---+---+---+
+---+---+---+---+---+---+---+---+
Byte 2 | Instrument | Command |
+---+---+---+---+---+---+---+---+
+---+---+---+---+---+---+---+---+
Byte 3 | Value |
+---+---+---+---+---+---+---+---+
Frequency is a 12 bit value ranging from $71 to $358. It is the value
put directly into the Amiga sound chip. The actual frequency can be
calculated as:
6
3.5 X 10 [3.579545 (PAL), 3.546895 (NTSC)]
f = -----------
value
This is an aproximate value. The actual value depends on the video
display and varies for PAL and NTSC. The exact values can be found in
the Amiga hardware reference manual. It is the horizontal scan line
frequency multiplied by 124 X 2, or something similar. A value of 0
means use the same as before.
Instrument is either a 4 or 5 bit value. If the extended instrument
is allowed thn the I bit becomes bit 4 of the instrument. A value of
0 means use the same instrument as before.
Command is one of:
0 : Arpeggio
1 : Portamento Up
2 : Portamento Down
3..9 : Reserved for future use
10 : Volume slide
11 : Position Jump
12 : Set Volume
13 : Pattern break
14 : Set filter
15 : Set speed
Value is used differently depending on the command:
Arpeggio:
A value of 0 means NO arpeggio. This also doubles as a NO COMMAND
instruction.
If the value is non-zero then it is split into 2 nybbles. The
note will warble between different frequencies of:
Base note
Base note + 1st nybble (semitones)
Base note + 2nd nybble (semitone)
Base note
Base note + 1st nybble (semitones)
Base note + 2nd nybble (semitone)
The note changes each VBL (20ms).
Portamento:
The value is either added or subtracted to the base note each VBL
depending on the direction. The value refers to the amiga
frequency as referred to above. The note is always forced into
the range $71..$358.
Volume slide:
The volume is changed by VALUE each VBL. Presumably it is in 2's
complement so that -ve and +ve values can be used. Presumably
values are clipped to the range 0..64?
Position Jump:
Value is the position within the Sequence list to jump to
Set Volume:
Set the current volume to value. This is a value between 0..64
and is the value used by the Amiga hardware!
Pattern Break:
Ends the current pattern.
Set Filter:
A value of 1 turns the Amiga's filter ON, value 0 turns it off.
Other values are undefined.
Set speed:
Value is the number of VBL's between events. The default is 6.
It should be a value between 1..15.