Files
Tharo 970af5600a Audio: Fix note values (#2716)
Due to the layout of the pitch_frequencies lookup table, note values
computed in extraction were reflected around middle C (midi note
number 60). This didn't matter for matching, values would successfully
roundtrip. However when using samples in external programs or
converting soundfonts to standard formats the note values would lead
to incorrect playback of sounds. This change corrects the note values
so that external programs correctly infer the pitch of the sound when
played at a particular MIDI key.
2026-03-10 06:22:28 +09:00

407 lines
15 KiB
Python

# SPDX-FileCopyrightText: © 2024 ZeldaRET
# SPDX-License-Identifier: CC0-1.0
#
# This file implements reading various structures resident to the Audiobank files.
# Additionally handles:
# - Linking with finalized samples
# - Writing xml elements representing these structures in soundfont xmls
#
import struct
from enum import IntEnum
from .audio_tables import AudioStorageMedium
from .tuning import rate_from_tuning, pitch_names
from .util import XMLWriter
VADPCM_VERSTAMP = 1
class AudioSampleCodec(IntEnum):
CODEC_ADPCM = 0
CODEC_S8 = 1
CODEC_S16_INMEMORY = 2
CODEC_SMALL_ADPCM = 3
CODEC_REVERB = 4
CODEC_S16 = 5
class SoundFontSample: # SampleHeader ?
"""
typedef struct {
/* 0x00 */ u32 codec : 4;
/* 0x00 */ u32 medium : 2; // storage medium determines which of the two sample bank ids to use when relocating sampleAddr
/* 0x00 */ u32 cached : 1;
/* 0x00 */ u32 isRelocated : 1;
/* 0x01 */ u32 size : 24;
/* 0x04 */ u8* sampleAddr; // offset into the sample bank associated with this soundfont
/* 0x08 */ AdpcmLoop* loop;
/* 0x0C */ AdpcmBook* book;
} SoundFontSample; // size = 0x10
"""
SIZE = 0x10
def __init__(self, data):
bits, self.sample_addr, self.loop, self.book = struct.unpack(">IIII", data[:0x10])
self.codec = AudioSampleCodec((bits >> 28) & 0b1111)
self.medium = AudioStorageMedium((bits >> 26) & 0b11)
self.cached = bool((bits >> 25) & 1)
self.is_relocated = bool((bits >> 24) & 1)
self.size = (bits >> 0) & 0b111111111111111111111111
assert self.book != 0
assert self.loop != 0
assert self.codec in [AudioSampleCodec.CODEC_ADPCM, AudioSampleCodec.CODEC_SMALL_ADPCM]
assert self.medium == 0
assert not self.is_relocated # Not relocated in ROM
def to_xml(self, xml : XMLWriter, name : str, rate_override = None, note_override = None):
# Example xml output:
# <Sample Name="SAMPLE_NAME" SampleRate="32000" BaseNote="C4" IsDD="false" Cached="false">
attrs = { "Name" : name }
if rate_override is not None:
attrs["SampleRate"] = rate_override
if note_override is not None:
attrs["BaseNote"] = pitch_names[note_override]
if self.medium != 0:
attrs["IsDD"] = "true"
if self.cached:
attrs["Cached"] = str(self.cached).lower()
xml.write_element("Sample", attrs)
def __str__(self):
out = "(SoundFontSample){\n"
out += f" .codec = {self.codec.name}\n"
out += f" .medium = {self.medium.name}\n"
out += f" .cached = {self.cached}\n"
out += f" .is_relocated = {self.is_relocated}\n"
out += f" .size = 0x{self.size:X}\n"
out += f" .sampleAddr = 0x{self.sample_addr:X}\n"
out += f" .loop = 0x{self.loop:X}\n"
out += f" .book = 0x{self.book:X}\n"
out += "}\n"
return out
class AdpcmLoop:
"""
typedef struct {
/* 0x00 */ u32 start;
/* 0x04 */ u32 end;
/* 0x08 */ u32 count;
/* 0x0C */ u32 numFrames;
/* 0x10 */ s16 state[16]; // only exists if count != 0. 8-byte aligned
} AdpcmLoop; // size = 0x30 (or 0x10)
"""
def __init__(self, data):
self.start, self.end, self.count, self.num_frames = struct.unpack(">IIII", data[:0x10])
# We expect loops to be either "no loop" or "infinite", as these are all that vadpcm_enc could handle.
assert self.count in (0,0xFFFFFFFF)
if self.count != 0:
self.state = tuple(s[0] for s in struct.iter_unpack(">h", data[0x10:0x30]))
else:
# A count of 0 indicates "no loop", but a loop structure is mandatory for all samples so something had to
# be emitted. Ensure the start is at 0, later we will ensure that the end is at the last frame of the sample
# once we have the sample data.
assert self.start == 0
self.state = tuple([0] * 16)
assert len(self.state) == 16
def serialize(self):
"""
Creates VADPCMLOOPS section data for aifc files
"""
NUM_LOOPS = 1
return struct.pack(">HHIII16h",
VADPCM_VERSTAMP, NUM_LOOPS,
self.start, self.end, self.count,
*self.state)
def __eq__(self, other):
if not isinstance(other, AdpcmLoop):
return False
other : AdpcmLoop
start_matches = self.start == other.start
end_matches = self.end == other.end
count_matches = self.count == other.count
# We don't check num_frames in loop equality since loops in different soundfonts referring to the same
# sample data may not have this field filled out
return start_matches and end_matches and count_matches and self.state == other.state
def __str__(self):
out = "(AdpcmLoop){\n"
out += f" .start = {self.start},\n"
out += f" .end = {self.end},\n"
out += f" .count = {self.count},\n"
out += f" .numFrames = {self.num_frames},\n"
out += f" .state = {self.state},\n"
out += "}\n"
return out
class AdpcmBook:
"""
typedef struct {
/* 0x00 */ s32 order;
/* 0x04 */ s32 npredictors;
/* 0x08 */ s16 book[1]; // size 8 * order * npredictors. 8-byte aligned
} AdpcmBook; // size >= 0x8
"""
def __init__(self, data):
self.order, self.n_predictors = struct.unpack(">ii", data[:8])
self.book = tuple(s[0] for s in struct.iter_unpack(">h", data[8:][:2 * 8 * self.order * self.n_predictors]))
assert len(self.book) == 8 * self.order * self.n_predictors , (len(self.book), 8 * self.order * self.n_predictors)
def serialize(self):
header = struct.pack(">hhh", VADPCM_VERSTAMP, self.order, self.n_predictors)
data = b"".join(struct.pack(">h", x) for x in self.book)
return header + data
def __eq__(self, other):
if not isinstance(other, AdpcmBook):
return False
other : AdpcmBook
order_matches = self.order == other.order
npredictors_matches = self.n_predictors == other.n_predictors
return order_matches and npredictors_matches and self.book == other.book
def __str__(self):
out = "(AdpcmBook){\n"
out += f" .order = {self.order},\n"
out += f" .npredictors = {self.n_predictors},\n"
out += f" .book = {self.book},\n"
out += "}\n"
return out
class SoundFontSound:
"""
typedef struct {
/* 0x00 */ SoundFontSample* sample;
/* 0x04 */ f32 tuning; // frequency scale factor
} SoundFontSound; // size = 0x8
"""
SIZE = 8
def __init__(self, index, data):
self.index = index
self.sample, self.tuning = struct.unpack(">If", data[:8])
def finalize(self, sample_lookup_fn):
from .audiotable import AudioTableSample
sample = sample_lookup_fn(self.sample)
if sample is None:
return
assert isinstance(sample, AudioTableSample)
sample : AudioTableSample
assert self.tuning in sample.tuning_map
rate,note = sample.tuning_map[self.tuning]
self.sample_rate = rate
self.needs_rate_override = self.sample_rate != sample.sample_rate
self.base_note = note
self.needs_note_override = self.base_note != sample.base_note
def __str__(self) -> str:
out = "(SoundFontSound}{\n"
out += f" .sample = 0x{self.sample:X}\n"
out += f" .tuning = {self.tuning:.7f}f\n"
out += "}\n"
return out
def to_xml(self, xml : XMLWriter, name : str, sample_name_func):
if self.sample == 0 and self.tuning == 0:
xml.write_element("Effect")
else:
attrs = {
"Name" : name,
"Sample" : sample_name_func(self.sample),
}
if self.needs_rate_override:
attrs["SampleRate"] = self.sample_rate
if self.needs_note_override:
attrs["BaseNote"] = pitch_names[self.base_note]
xml.write_element("Effect", attrs)
class Drum:
"""
typedef struct {
/* 0x00 */ u8 releaseRate;
/* 0x01 */ u8 pan;
/* 0x02 */ u8 isRelocated;
/* 0x04 */ SoundFontSound sound;
/* 0x0C */ AdsrEnvelope* envelope;
} Drum; // size = 0x10
"""
SIZE = 0x10
def __init__(self, data):
self.release_rate, self.pan, self.is_relocated, self.sample, self.tuning, self.envelope = \
struct.unpack(">BBBxIfI", data[:0x10])
assert self.is_relocated == 0
def group_continuation(self, other):
"""
Determine if self is a continuation of the drum group containing other, the last drum added.
"""
# If there is no previous drum or the previous drum was an empty entry, always begin a new group
if other is None:
return False
assert isinstance(other, Drum)
# Check general agreement, if these attributes do not match it is certainly not part of the same group
if self.sample == other.sample and self.pan == other.pan and self.envelope == other.envelope and \
self.release_rate == other.release_rate:
# If there is any intersection in the samplerates, assume these are in the same drum group
samplerates1 = set(rate for _,rate in rate_from_tuning(self.tuning))
samplerates2 = set(rate for _,rate in rate_from_tuning(other.tuning))
return len(samplerates1.intersection(samplerates2)) != 0
return False
def __str__(self):
out = "(Drum){\n"
out += f" .releaseRate = {self.release_rate},\n"
out += f" .pan = {self.pan},\n"
out += f" .isRelocated = {self.is_relocated},\n"
out += f" .sound.sample = 0x{self.sample:X},\n"
out += f" .sound.tuning = {self.tuning:.7f}f,\n"
out += f" .envelope = 0x{self.envelope:X},\n"
out += "}\n"
return out
class Instrument:
"""
typedef struct {
/* 0x00 */ u8 isRelocated;
/* 0x01 */ u8 normalRangeLo;
/* 0x02 */ u8 normalRangeHi;
/* 0x03 */ u8 releaseRate;
/* 0x04 */ AdsrEnvelope* envelope;
/* 0x08 */ SoundFontSound lowNotesSound;
/* 0x10 */ SoundFontSound normalNotesSound;
/* 0x18 */ SoundFontSound highNotesSound;
} Instrument; // size = 0x20
"""
SIZE = 0x20
def __init__(self, data):
self.is_relocated, self.normal_range_lo, self.normal_range_hi, self.release_rate, self.envelope, \
self.low_notes_sample, self.low_notes_tuning, \
self.normal_notes_sample, self.normal_notes_tuning, \
self.high_notes_sample, self.high_notes_tuning = struct.unpack(">BBBBIIfIfIf", data[:0x20])
self.program_number = None
self.offset = None
self.struct_index = None
self.unused = False
assert self.is_relocated == 0
# Sample is either present or the split point is at the start/end
assert not (self.low_notes_sample == 0 and self.low_notes_tuning == 0.0) or self.normal_range_lo == 0
assert not (self.high_notes_sample == 0 and self.high_notes_tuning == 0.0) or self.normal_range_hi == 127
def __str__(self):
out = "(Instrument){\n"
out += f" .isRelocated = {self.is_relocated},\n"
out += f" .normalRangeLo = {self.normal_range_lo},\n"
out += f" .normalRangeHi = {self.normal_range_hi},\n"
out += f" .releaseRate = {self.release_rate},\n"
out += f" .envelope = 0x{self.envelope:X},\n"
out += f" .lowNotesSound.sample = {self.low_notes_sample},\n"
out += f" .lowNotesSound.tuning = {self.low_notes_tuning},\n"
out += f" .normalNotesSound.sample = {self.normal_notes_sample},\n"
out += f" .normalNotesSound.tuning = {self.normal_notes_tuning},\n"
out += f" .highNotesSound.sample = {self.high_notes_sample},\n"
out += f" .highNotesSound.tuning = {self.high_notes_tuning},\n"
out += "}\n"
return out
def finalize(self, sample_lookup_fn):
from .audiotable import AudioTableSample
self.sample_rate = [None] * 3
self.base_note = [None] * 3
self.needs_rate_override = [False] * 3
self.needs_note_override = [False] * 3
sample_offsets = (self.low_notes_sample, self.normal_notes_sample, self.high_notes_sample)
tunings = (self.low_notes_tuning, self.normal_notes_tuning, self.high_notes_tuning)
for i,(sample_offset,tuning) in enumerate(zip(sample_offsets, tunings)):
sample = sample_lookup_fn(sample_offset)
if sample is None:
continue
assert isinstance(sample, AudioTableSample)
sample : AudioTableSample
assert tuning in sample.tuning_map
rate,note = sample.tuning_map[tuning]
self.sample_rate[i] = rate
self.needs_rate_override[i] = self.sample_rate[i] != sample.sample_rate
self.base_note[i] = note
self.needs_note_override[i] = self.base_note[i] != sample.base_note
def to_xml(self, xml : XMLWriter, name : str, sample_names_func, envelope_name_func):
attributes = {}
if not self.unused:
attributes["ProgramNumber"] = self.program_number
attributes["Name"] = name
# TODO release rate overrides?
attributes.update({
"Envelope" : envelope_name_func(self.envelope),
#"Release" : self.release_rate,
"Sample" : sample_names_func(self.normal_notes_sample),
})
if self.needs_rate_override[1]:
attributes["SampleRate"] = self.sample_rate[1]
if self.needs_note_override[1]:
attributes["BaseNote"] = pitch_names[self.base_note[1]]
if self.normal_range_lo != 0:
attributes["RangeLo"] = pitch_names[self.normal_range_lo]
attributes["SampleLo"] = sample_names_func(self.low_notes_sample)
if self.needs_rate_override[0]:
attributes["SampleRateLo"] = self.sample_rate[0]
if self.needs_note_override[0]:
attributes["BaseNoteLo"] = pitch_names[self.base_note[0]]
if self.normal_range_hi != 127:
attributes["RangeHi"] = pitch_names[self.normal_range_hi]
attributes["SampleHi"] = sample_names_func(self.high_notes_sample)
if self.needs_rate_override[2]:
attributes["SampleRateHi"] = self.sample_rate[2]
if self.needs_note_override[2]:
attributes["BaseNoteHi"] = pitch_names[self.base_note[2]]
xml.write_element("Instrument" if not self.unused else "InstrumentUnused", attributes)