godot/scene/resources/audio_stream_wav.cpp
2024-02-13 15:37:09 +01:00

668 lines
19 KiB
C++

/**************************************************************************/
/* audio_stream_wav.cpp */
/**************************************************************************/
/* This file is part of: */
/* GODOT ENGINE */
/* https://godotengine.org */
/**************************************************************************/
/* Copyright (c) 2014-present Godot Engine contributors (see AUTHORS.md). */
/* Copyright (c) 2007-2014 Juan Linietsky, Ariel Manzur. */
/* */
/* Permission is hereby granted, free of charge, to any person obtaining */
/* a copy of this software and associated documentation files (the */
/* "Software"), to deal in the Software without restriction, including */
/* without limitation the rights to use, copy, modify, merge, publish, */
/* distribute, sublicense, and/or sell copies of the Software, and to */
/* permit persons to whom the Software is furnished to do so, subject to */
/* the following conditions: */
/* */
/* The above copyright notice and this permission notice shall be */
/* included in all copies or substantial portions of the Software. */
/* */
/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, */
/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF */
/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. */
/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY */
/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, */
/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE */
/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */
/**************************************************************************/
#include "audio_stream_wav.h"
#include "core/io/file_access.h"
#include "core/io/marshalls.h"
void AudioStreamPlaybackWAV::start(double p_from_pos) {
if (base->format == AudioStreamWAV::FORMAT_IMA_ADPCM) {
//no seeking in IMA_ADPCM
for (int i = 0; i < 2; i++) {
ima_adpcm[i].step_index = 0;
ima_adpcm[i].predictor = 0;
ima_adpcm[i].loop_step_index = 0;
ima_adpcm[i].loop_predictor = 0;
ima_adpcm[i].last_nibble = -1;
ima_adpcm[i].loop_pos = 0x7FFFFFFF;
ima_adpcm[i].window_ofs = 0;
}
offset = 0;
} else {
seek(p_from_pos);
}
sign = 1;
active = true;
}
void AudioStreamPlaybackWAV::stop() {
active = false;
}
bool AudioStreamPlaybackWAV::is_playing() const {
return active;
}
int AudioStreamPlaybackWAV::get_loop_count() const {
return 0;
}
double AudioStreamPlaybackWAV::get_playback_position() const {
return float(offset >> MIX_FRAC_BITS) / base->mix_rate;
}
void AudioStreamPlaybackWAV::seek(double p_time) {
if (base->format == AudioStreamWAV::FORMAT_IMA_ADPCM) {
return; //no seeking in ima-adpcm
}
double max = base->get_length();
if (p_time < 0) {
p_time = 0;
} else if (p_time >= max) {
p_time = max - 0.001;
}
offset = uint64_t(p_time * base->mix_rate) << MIX_FRAC_BITS;
}
template <class Depth, bool is_stereo, bool is_ima_adpcm>
void AudioStreamPlaybackWAV::do_resample(const Depth *p_src, AudioFrame *p_dst, int64_t &p_offset, int32_t &p_increment, uint32_t p_amount, IMA_ADPCM_State *p_ima_adpcm) {
// this function will be compiled branchless by any decent compiler
int32_t final, final_r, next, next_r;
while (p_amount) {
p_amount--;
int64_t pos = p_offset >> MIX_FRAC_BITS;
if (is_stereo && !is_ima_adpcm) {
pos <<= 1;
}
if (is_ima_adpcm) {
int64_t sample_pos = pos + p_ima_adpcm[0].window_ofs;
while (sample_pos > p_ima_adpcm[0].last_nibble) {
static const int16_t _ima_adpcm_step_table[89] = {
7, 8, 9, 10, 11, 12, 13, 14, 16, 17,
19, 21, 23, 25, 28, 31, 34, 37, 41, 45,
50, 55, 60, 66, 73, 80, 88, 97, 107, 118,
130, 143, 157, 173, 190, 209, 230, 253, 279, 307,
337, 371, 408, 449, 494, 544, 598, 658, 724, 796,
876, 963, 1060, 1166, 1282, 1411, 1552, 1707, 1878, 2066,
2272, 2499, 2749, 3024, 3327, 3660, 4026, 4428, 4871, 5358,
5894, 6484, 7132, 7845, 8630, 9493, 10442, 11487, 12635, 13899,
15289, 16818, 18500, 20350, 22385, 24623, 27086, 29794, 32767
};
static const int8_t _ima_adpcm_index_table[16] = {
-1, -1, -1, -1, 2, 4, 6, 8,
-1, -1, -1, -1, 2, 4, 6, 8
};
for (int i = 0; i < (is_stereo ? 2 : 1); i++) {
int16_t nibble, diff, step;
p_ima_adpcm[i].last_nibble++;
const uint8_t *src_ptr = (const uint8_t *)base->data;
src_ptr += AudioStreamWAV::DATA_PAD;
uint8_t nbb = src_ptr[(p_ima_adpcm[i].last_nibble >> 1) * (is_stereo ? 2 : 1) + i];
nibble = (p_ima_adpcm[i].last_nibble & 1) ? (nbb >> 4) : (nbb & 0xF);
step = _ima_adpcm_step_table[p_ima_adpcm[i].step_index];
p_ima_adpcm[i].step_index += _ima_adpcm_index_table[nibble];
if (p_ima_adpcm[i].step_index < 0) {
p_ima_adpcm[i].step_index = 0;
}
if (p_ima_adpcm[i].step_index > 88) {
p_ima_adpcm[i].step_index = 88;
}
diff = step >> 3;
if (nibble & 1) {
diff += step >> 2;
}
if (nibble & 2) {
diff += step >> 1;
}
if (nibble & 4) {
diff += step;
}
if (nibble & 8) {
diff = -diff;
}
p_ima_adpcm[i].predictor += diff;
if (p_ima_adpcm[i].predictor < -0x8000) {
p_ima_adpcm[i].predictor = -0x8000;
} else if (p_ima_adpcm[i].predictor > 0x7FFF) {
p_ima_adpcm[i].predictor = 0x7FFF;
}
/* store loop if there */
if (p_ima_adpcm[i].last_nibble == p_ima_adpcm[i].loop_pos) {
p_ima_adpcm[i].loop_step_index = p_ima_adpcm[i].step_index;
p_ima_adpcm[i].loop_predictor = p_ima_adpcm[i].predictor;
}
//printf("%i - %i - pred %i\n",int(p_ima_adpcm[i].last_nibble),int(nibble),int(p_ima_adpcm[i].predictor));
}
}
final = p_ima_adpcm[0].predictor;
if (is_stereo) {
final_r = p_ima_adpcm[1].predictor;
}
} else {
final = p_src[pos];
if (is_stereo) {
final_r = p_src[pos + 1];
}
if constexpr (sizeof(Depth) == 1) { /* conditions will not exist anymore when compiled! */
final <<= 8;
if (is_stereo) {
final_r <<= 8;
}
}
if (is_stereo) {
next = p_src[pos + 2];
next_r = p_src[pos + 3];
} else {
next = p_src[pos + 1];
}
if constexpr (sizeof(Depth) == 1) {
next <<= 8;
if (is_stereo) {
next_r <<= 8;
}
}
int32_t frac = int64_t(p_offset & MIX_FRAC_MASK);
final = final + ((next - final) * frac >> MIX_FRAC_BITS);
if (is_stereo) {
final_r = final_r + ((next_r - final_r) * frac >> MIX_FRAC_BITS);
}
}
if (!is_stereo) {
final_r = final; //copy to right channel if stereo
}
p_dst->left = final / 32767.0;
p_dst->right = final_r / 32767.0;
p_dst++;
p_offset += p_increment;
}
}
int AudioStreamPlaybackWAV::mix(AudioFrame *p_buffer, float p_rate_scale, int p_frames) {
if (!base->data || !active) {
for (int i = 0; i < p_frames; i++) {
p_buffer[i] = AudioFrame(0, 0);
}
return 0;
}
int len = base->data_bytes;
switch (base->format) {
case AudioStreamWAV::FORMAT_8_BITS:
len /= 1;
break;
case AudioStreamWAV::FORMAT_16_BITS:
len /= 2;
break;
case AudioStreamWAV::FORMAT_IMA_ADPCM:
len *= 2;
break;
}
if (base->stereo) {
len /= 2;
}
/* some 64-bit fixed point precaches */
int64_t loop_begin_fp = ((int64_t)base->loop_begin << MIX_FRAC_BITS);
int64_t loop_end_fp = ((int64_t)base->loop_end << MIX_FRAC_BITS);
int64_t length_fp = ((int64_t)len << MIX_FRAC_BITS);
int64_t begin_limit = (base->loop_mode != AudioStreamWAV::LOOP_DISABLED) ? loop_begin_fp : 0;
int64_t end_limit = (base->loop_mode != AudioStreamWAV::LOOP_DISABLED) ? loop_end_fp : length_fp;
bool is_stereo = base->stereo;
int32_t todo = p_frames;
if (base->loop_mode == AudioStreamWAV::LOOP_BACKWARD) {
sign = -1;
}
float base_rate = AudioServer::get_singleton()->get_mix_rate();
float srate = base->mix_rate;
srate *= p_rate_scale;
float playback_speed_scale = AudioServer::get_singleton()->get_playback_speed_scale();
float fincrement = (srate * playback_speed_scale) / base_rate;
int32_t increment = int32_t(MAX(fincrement * MIX_FRAC_LEN, 1));
increment *= sign;
//looping
AudioStreamWAV::LoopMode loop_format = base->loop_mode;
AudioStreamWAV::Format format = base->format;
/* audio data */
uint8_t *dataptr = (uint8_t *)base->data;
const void *data = dataptr + AudioStreamWAV::DATA_PAD;
AudioFrame *dst_buff = p_buffer;
if (format == AudioStreamWAV::FORMAT_IMA_ADPCM) {
if (loop_format != AudioStreamWAV::LOOP_DISABLED) {
ima_adpcm[0].loop_pos = loop_begin_fp >> MIX_FRAC_BITS;
ima_adpcm[1].loop_pos = loop_begin_fp >> MIX_FRAC_BITS;
loop_format = AudioStreamWAV::LOOP_FORWARD;
}
}
while (todo > 0) {
int64_t limit = 0;
int32_t target = 0, aux = 0;
/** LOOP CHECKING **/
if (increment < 0) {
/* going backwards */
if (loop_format != AudioStreamWAV::LOOP_DISABLED && offset < loop_begin_fp) {
/* loopstart reached */
if (loop_format == AudioStreamWAV::LOOP_PINGPONG) {
/* bounce ping pong */
offset = loop_begin_fp + (loop_begin_fp - offset);
increment = -increment;
sign *= -1;
} else {
/* go to loop-end */
offset = loop_end_fp - (loop_begin_fp - offset);
}
} else {
/* check for sample not reaching beginning */
if (offset < 0) {
active = false;
break;
}
}
} else {
/* going forward */
if (loop_format != AudioStreamWAV::LOOP_DISABLED && offset >= loop_end_fp) {
/* loopend reached */
if (loop_format == AudioStreamWAV::LOOP_PINGPONG) {
/* bounce ping pong */
offset = loop_end_fp - (offset - loop_end_fp);
increment = -increment;
sign *= -1;
} else {
/* go to loop-begin */
if (format == AudioStreamWAV::FORMAT_IMA_ADPCM) {
for (int i = 0; i < 2; i++) {
ima_adpcm[i].step_index = ima_adpcm[i].loop_step_index;
ima_adpcm[i].predictor = ima_adpcm[i].loop_predictor;
ima_adpcm[i].last_nibble = loop_begin_fp >> MIX_FRAC_BITS;
}
offset = loop_begin_fp;
} else {
offset = loop_begin_fp + (offset - loop_end_fp);
}
}
} else {
/* no loop, check for end of sample */
if (offset >= length_fp) {
active = false;
break;
}
}
}
/** MIXCOUNT COMPUTING **/
/* next possible limit (looppoints or sample begin/end */
limit = (increment < 0) ? begin_limit : end_limit;
/* compute what is shorter, the todo or the limit? */
aux = (limit - offset) / increment + 1;
target = (aux < todo) ? aux : todo; /* mix target is the shorter buffer */
/* check just in case */
if (target <= 0) {
active = false;
break;
}
todo -= target;
switch (base->format) {
case AudioStreamWAV::FORMAT_8_BITS: {
if (is_stereo) {
do_resample<int8_t, true, false>((int8_t *)data, dst_buff, offset, increment, target, ima_adpcm);
} else {
do_resample<int8_t, false, false>((int8_t *)data, dst_buff, offset, increment, target, ima_adpcm);
}
} break;
case AudioStreamWAV::FORMAT_16_BITS: {
if (is_stereo) {
do_resample<int16_t, true, false>((int16_t *)data, dst_buff, offset, increment, target, ima_adpcm);
} else {
do_resample<int16_t, false, false>((int16_t *)data, dst_buff, offset, increment, target, ima_adpcm);
}
} break;
case AudioStreamWAV::FORMAT_IMA_ADPCM: {
if (is_stereo) {
do_resample<int8_t, true, true>((int8_t *)data, dst_buff, offset, increment, target, ima_adpcm);
} else {
do_resample<int8_t, false, true>((int8_t *)data, dst_buff, offset, increment, target, ima_adpcm);
}
} break;
}
dst_buff += target;
}
if (todo) {
int mixed_frames = p_frames - todo;
//bit was missing from mix
int todo_ofs = p_frames - todo;
for (int i = todo_ofs; i < p_frames; i++) {
p_buffer[i] = AudioFrame(0, 0);
}
return mixed_frames;
}
return p_frames;
}
void AudioStreamPlaybackWAV::tag_used_streams() {
base->tag_used(get_playback_position());
}
AudioStreamPlaybackWAV::AudioStreamPlaybackWAV() {}
/////////////////////
void AudioStreamWAV::set_format(Format p_format) {
format = p_format;
}
AudioStreamWAV::Format AudioStreamWAV::get_format() const {
return format;
}
void AudioStreamWAV::set_loop_mode(LoopMode p_loop_mode) {
loop_mode = p_loop_mode;
}
AudioStreamWAV::LoopMode AudioStreamWAV::get_loop_mode() const {
return loop_mode;
}
void AudioStreamWAV::set_loop_begin(int p_frame) {
loop_begin = p_frame;
}
int AudioStreamWAV::get_loop_begin() const {
return loop_begin;
}
void AudioStreamWAV::set_loop_end(int p_frame) {
loop_end = p_frame;
}
int AudioStreamWAV::get_loop_end() const {
return loop_end;
}
void AudioStreamWAV::set_mix_rate(int p_hz) {
ERR_FAIL_COND(p_hz == 0);
mix_rate = p_hz;
}
int AudioStreamWAV::get_mix_rate() const {
return mix_rate;
}
void AudioStreamWAV::set_stereo(bool p_enable) {
stereo = p_enable;
}
bool AudioStreamWAV::is_stereo() const {
return stereo;
}
double AudioStreamWAV::get_length() const {
int len = data_bytes;
switch (format) {
case AudioStreamWAV::FORMAT_8_BITS:
len /= 1;
break;
case AudioStreamWAV::FORMAT_16_BITS:
len /= 2;
break;
case AudioStreamWAV::FORMAT_IMA_ADPCM:
len *= 2;
break;
}
if (stereo) {
len /= 2;
}
return double(len) / mix_rate;
}
bool AudioStreamWAV::is_monophonic() const {
return false;
}
void AudioStreamWAV::set_data(const Vector<uint8_t> &p_data) {
AudioServer::get_singleton()->lock();
if (data) {
memfree(data);
data = nullptr;
data_bytes = 0;
}
int datalen = p_data.size();
if (datalen) {
const uint8_t *r = p_data.ptr();
int alloc_len = datalen + DATA_PAD * 2;
data = memalloc(alloc_len); //alloc with some padding for interpolation
memset(data, 0, alloc_len);
uint8_t *dataptr = (uint8_t *)data;
memcpy(dataptr + DATA_PAD, r, datalen);
data_bytes = datalen;
}
AudioServer::get_singleton()->unlock();
}
Vector<uint8_t> AudioStreamWAV::get_data() const {
Vector<uint8_t> pv;
if (data) {
pv.resize(data_bytes);
{
uint8_t *w = pv.ptrw();
uint8_t *dataptr = (uint8_t *)data;
memcpy(w, dataptr + DATA_PAD, data_bytes);
}
}
return pv;
}
Error AudioStreamWAV::save_to_wav(const String &p_path) {
if (format == AudioStreamWAV::FORMAT_IMA_ADPCM) {
WARN_PRINT("Saving IMA_ADPC samples are not supported yet");
return ERR_UNAVAILABLE;
}
int sub_chunk_2_size = data_bytes; //Subchunk2Size = Size of data in bytes
// Format code
// 1:PCM format (for 8 or 16 bit)
// 3:IEEE float format
int format_code = (format == FORMAT_IMA_ADPCM) ? 3 : 1;
int n_channels = stereo ? 2 : 1;
long sample_rate = mix_rate;
int byte_pr_sample = 0;
switch (format) {
case AudioStreamWAV::FORMAT_8_BITS:
byte_pr_sample = 1;
break;
case AudioStreamWAV::FORMAT_16_BITS:
byte_pr_sample = 2;
break;
case AudioStreamWAV::FORMAT_IMA_ADPCM:
byte_pr_sample = 4;
break;
}
String file_path = p_path;
if (!(file_path.substr(file_path.length() - 4, 4) == ".wav")) {
file_path += ".wav";
}
Ref<FileAccess> file = FileAccess::open(file_path, FileAccess::WRITE); //Overrides existing file if present
ERR_FAIL_COND_V(file.is_null(), ERR_FILE_CANT_WRITE);
// Create WAV Header
file->store_string("RIFF"); //ChunkID
file->store_32(sub_chunk_2_size + 36); //ChunkSize = 36 + SubChunk2Size (size of entire file minus the 8 bits for this and previous header)
file->store_string("WAVE"); //Format
file->store_string("fmt "); //Subchunk1ID
file->store_32(16); //Subchunk1Size = 16
file->store_16(format_code); //AudioFormat
file->store_16(n_channels); //Number of Channels
file->store_32(sample_rate); //SampleRate
file->store_32(sample_rate * n_channels * byte_pr_sample); //ByteRate
file->store_16(n_channels * byte_pr_sample); //BlockAlign = NumChannels * BytePrSample
file->store_16(byte_pr_sample * 8); //BitsPerSample
file->store_string("data"); //Subchunk2ID
file->store_32(sub_chunk_2_size); //Subchunk2Size
// Add data
Vector<uint8_t> stream_data = get_data();
const uint8_t *read_data = stream_data.ptr();
switch (format) {
case AudioStreamWAV::FORMAT_8_BITS:
for (unsigned int i = 0; i < data_bytes; i++) {
uint8_t data_point = (read_data[i] + 128);
file->store_8(data_point);
}
break;
case AudioStreamWAV::FORMAT_16_BITS:
for (unsigned int i = 0; i < data_bytes / 2; i++) {
uint16_t data_point = decode_uint16(&read_data[i * 2]);
file->store_16(data_point);
}
break;
case AudioStreamWAV::FORMAT_IMA_ADPCM:
//Unimplemented
break;
}
return OK;
}
Ref<AudioStreamPlayback> AudioStreamWAV::instantiate_playback() {
Ref<AudioStreamPlaybackWAV> sample;
sample.instantiate();
sample->base = Ref<AudioStreamWAV>(this);
return sample;
}
String AudioStreamWAV::get_stream_name() const {
return "";
}
void AudioStreamWAV::_bind_methods() {
ClassDB::bind_method(D_METHOD("set_data", "data"), &AudioStreamWAV::set_data);
ClassDB::bind_method(D_METHOD("get_data"), &AudioStreamWAV::get_data);
ClassDB::bind_method(D_METHOD("set_format", "format"), &AudioStreamWAV::set_format);
ClassDB::bind_method(D_METHOD("get_format"), &AudioStreamWAV::get_format);
ClassDB::bind_method(D_METHOD("set_loop_mode", "loop_mode"), &AudioStreamWAV::set_loop_mode);
ClassDB::bind_method(D_METHOD("get_loop_mode"), &AudioStreamWAV::get_loop_mode);
ClassDB::bind_method(D_METHOD("set_loop_begin", "loop_begin"), &AudioStreamWAV::set_loop_begin);
ClassDB::bind_method(D_METHOD("get_loop_begin"), &AudioStreamWAV::get_loop_begin);
ClassDB::bind_method(D_METHOD("set_loop_end", "loop_end"), &AudioStreamWAV::set_loop_end);
ClassDB::bind_method(D_METHOD("get_loop_end"), &AudioStreamWAV::get_loop_end);
ClassDB::bind_method(D_METHOD("set_mix_rate", "mix_rate"), &AudioStreamWAV::set_mix_rate);
ClassDB::bind_method(D_METHOD("get_mix_rate"), &AudioStreamWAV::get_mix_rate);
ClassDB::bind_method(D_METHOD("set_stereo", "stereo"), &AudioStreamWAV::set_stereo);
ClassDB::bind_method(D_METHOD("is_stereo"), &AudioStreamWAV::is_stereo);
ClassDB::bind_method(D_METHOD("save_to_wav", "path"), &AudioStreamWAV::save_to_wav);
ADD_PROPERTY(PropertyInfo(Variant::PACKED_BYTE_ARRAY, "data", PROPERTY_HINT_NONE, "", PROPERTY_USAGE_NO_EDITOR), "set_data", "get_data");
ADD_PROPERTY(PropertyInfo(Variant::INT, "format", PROPERTY_HINT_ENUM, "8-Bit,16-Bit,IMA-ADPCM"), "set_format", "get_format");
ADD_PROPERTY(PropertyInfo(Variant::INT, "loop_mode", PROPERTY_HINT_ENUM, "Disabled,Forward,Ping-Pong,Backward"), "set_loop_mode", "get_loop_mode");
ADD_PROPERTY(PropertyInfo(Variant::INT, "loop_begin"), "set_loop_begin", "get_loop_begin");
ADD_PROPERTY(PropertyInfo(Variant::INT, "loop_end"), "set_loop_end", "get_loop_end");
ADD_PROPERTY(PropertyInfo(Variant::INT, "mix_rate"), "set_mix_rate", "get_mix_rate");
ADD_PROPERTY(PropertyInfo(Variant::BOOL, "stereo"), "set_stereo", "is_stereo");
BIND_ENUM_CONSTANT(FORMAT_8_BITS);
BIND_ENUM_CONSTANT(FORMAT_16_BITS);
BIND_ENUM_CONSTANT(FORMAT_IMA_ADPCM);
BIND_ENUM_CONSTANT(LOOP_DISABLED);
BIND_ENUM_CONSTANT(LOOP_FORWARD);
BIND_ENUM_CONSTANT(LOOP_PINGPONG);
BIND_ENUM_CONSTANT(LOOP_BACKWARD);
}
AudioStreamWAV::AudioStreamWAV() {}
AudioStreamWAV::~AudioStreamWAV() {
if (data) {
memfree(data);
data = nullptr;
data_bytes = 0;
}
}