BREAKING: Re-port MIDI, faster, lower memory, ESP32 support (#790)
* Pull untouched upstream TinySoundFont Start from the upstream TinySoundFont by @schellingb at https://github.com/schellingb/TinySoundFont and rearchitect the MIDI processing. * Convert to PROGMEM-based SoundFont structures. This makes it much faster to access the SoundFont and also removes all the static RAM usage. The samples, the regions, the instruments are all now in ROM. Only the dynamic voices and related info takes up RAM. No caching. * (Re)convert to fixed point math. TSF upstream is built for systems with real floating point units (and produces great sound with it!), but processors like the Pico, ESP8266, and several models of ESP32 don't have one so emulate FP in *extremely* slow SW. Use fixed point (integer) representations for the inner loop and wherever possible to avoid `float` math. For processors like the Pico 2 or original ESP32 with hardware FPUs this still increases performance since integer is still much faster. * Redo the example and add default SF2s Rewrite the MIDI examples to use PROGMEM for both SF2 and MID file. It simplifies life for the casual user and me. * Breaking change The MIDI object now has a `midi::SetSoundFont(tsf *)` instead of a `midi::SetSoundfont(AudioFileSource *)`. It no longer supports reading SF2 from LittleFS or SD.
This commit is contained in:
1 parent
3e099fefe1
commit
7d99e6b7b9
31 files changed
+64533
-2836
No files matched your search
@@ -21,9 +21,15 @@
|
||||
#ifndef _AUDIOFILESOURCESPIFFS_H
|
||||
#define _AUDIOFILESOURCESPIFFS_H
|
||||
|
||||
#ifndef ARDUINO_ARCH_RP2040
|
||||
|
||||
#include <Arduino.h>
|
||||
#include <FS.h>
|
||||
|
||||
#ifdef ESP32
|
||||
#include <SPIFFS.h>
|
||||
#endif
|
||||
|
||||
#include "AudioFileSource.h"
|
||||
#include "AudioFileSourceFS.h"
|
||||
|
||||
@@ -37,6 +43,7 @@ public:
|
||||
// Others are inherited from base
|
||||
};
|
||||
|
||||
#endif
|
||||
|
||||
#endif
|
||||
|
||||
+53
-124
@@ -58,38 +58,19 @@
|
||||
|
||||
#include "AudioGeneratorMIDI.h"
|
||||
|
||||
#if defined(ESP32) && (__GNUC__ >= 8) && (__XTENSA__)
|
||||
// Do not build, Espressif's GCC8+ has a compiler bug
|
||||
#else // __GNUC__ == 8
|
||||
|
||||
#pragma GCC optimize ("O3")
|
||||
|
||||
#define TSF_NO_STDIO
|
||||
#define TSF_CONST_FILE
|
||||
#define TSF_SAMPLES_SHORT
|
||||
#define TSF_IMPLEMENTATION
|
||||
#include "libtinysoundfont/tsf.h"
|
||||
|
||||
|
||||
/**************** utility routines **********************/
|
||||
|
||||
/* announce a fatal MIDI file format error */
|
||||
|
||||
void AudioGeneratorMIDI::midi_error(const char *msg, int curpos) {
|
||||
cb.st(curpos, msg);
|
||||
#if 0
|
||||
int ptr;
|
||||
audioLogger->printf("---> MIDI file error at position %04X (%d): %s\n", (uint16_t) curpos, (uint16_t) curpos, msg);
|
||||
/* print some bytes surrounding the error */
|
||||
ptr = curpos - 16;
|
||||
if (ptr < 0) {
|
||||
ptr = 0;
|
||||
}
|
||||
buffer.seek(buffer.data, ptr);
|
||||
for (int i = 0; i < 32; i++) {
|
||||
char c;
|
||||
buffer.read(buffer.data, &c, 1);
|
||||
audioLogger->printf((ptr + i) == curpos ? " [%02X] " : "%02X ", (int) c & 0xff);
|
||||
}
|
||||
audioLogger->printf("\n");
|
||||
#endif
|
||||
running = false;
|
||||
}
|
||||
|
||||
@@ -119,8 +100,8 @@ void AudioGeneratorMIDI::process_header(void) {
|
||||
unsigned int time_division;
|
||||
|
||||
chk_bufdata(hdrptr, sizeof(struct midi_header));
|
||||
buffer.seek(buffer.data, hdrptr);
|
||||
buffer.read(buffer.data, &hdr, sizeof(hdr));
|
||||
file->seek(hdrptr, SEEK_SET);
|
||||
file->read(&hdr, sizeof(hdr));
|
||||
if (!charcmp((char *) hdr.MThd, "MThd")) {
|
||||
midi_error("Missing 'MThd'", hdrptr);
|
||||
}
|
||||
@@ -143,8 +124,10 @@ void AudioGeneratorMIDI::start_track(int tracknum) {
|
||||
unsigned long tracklen;
|
||||
|
||||
chk_bufdata(hdrptr, sizeof(struct track_header));
|
||||
buffer.seek(buffer.data, hdrptr);
|
||||
buffer.read(buffer.data, &hdr, sizeof(hdr));
|
||||
file->seek(hdrptr, SEEK_SET);
|
||||
//buffer.seek(buffer.data, hdrptr);
|
||||
file->read(&hdr, sizeof(hdr));
|
||||
//buffer.read(buffer.data, &hdr, sizeof(hdr));
|
||||
if (!charcmp((char *)(hdr.MTrk), "MTrk")) {
|
||||
midi_error("Missing 'MTrk'", hdrptr);
|
||||
}
|
||||
@@ -158,22 +141,28 @@ void AudioGeneratorMIDI::start_track(int tracknum) {
|
||||
|
||||
unsigned char AudioGeneratorMIDI::buffer_byte(int offset) {
|
||||
unsigned char c;
|
||||
buffer.seek(buffer.data, offset);
|
||||
buffer.read(buffer.data, &c, 1);
|
||||
file->seek(offset, SEEK_SET);
|
||||
//buffer.seek(buffer.data, offset);
|
||||
file->read(&c, 1);
|
||||
//buffer.read(buffer.data, &c, 1);
|
||||
return c;
|
||||
}
|
||||
|
||||
unsigned short AudioGeneratorMIDI::buffer_short(int offset) {
|
||||
unsigned short s;
|
||||
buffer.seek(buffer.data, offset);
|
||||
buffer.read(buffer.data, &s, sizeof(short));
|
||||
file->seek(offset, SEEK_SET);
|
||||
//buffer.seek(buffer.data, offset);
|
||||
file->read(&s, sizeof(short));
|
||||
// buffer.read(buffer.data, &s, sizeof(short));
|
||||
return s;
|
||||
}
|
||||
|
||||
unsigned int AudioGeneratorMIDI::buffer_int32(int offset) {
|
||||
uint32_t i;
|
||||
buffer.seek(buffer.data, offset);
|
||||
buffer.read(buffer.data, &i, sizeof(i));
|
||||
file->seek(offset, SEEK_SET);
|
||||
//buffer.seek(buffer.data, offset);
|
||||
file->read(&i, sizeof(i));
|
||||
//buffer.read(buffer.data, &i, sizeof(i));
|
||||
return i;
|
||||
}
|
||||
|
||||
@@ -345,18 +334,15 @@ note_off:
|
||||
|
||||
|
||||
// Open file, parse headers, get ready to process MIDI
|
||||
void AudioGeneratorMIDI::PrepareMIDI(AudioFileSource *src) {
|
||||
MakeStreamFromAFS(src, &afsMIDI);
|
||||
tsf_stream_wrap_cached(&afsMIDI, 32, 64, &buffer);
|
||||
buflen = buffer.size(buffer.data);
|
||||
void AudioGeneratorMIDI::PrepareMIDI() { //AudioFileSource *src) {
|
||||
buflen = file->getSize(); //buffer.size(buffer.data);
|
||||
|
||||
/* process the MIDI file header */
|
||||
|
||||
hdrptr = buffer.tell(buffer.data); /* pointer to file and track headers */
|
||||
hdrptr = file->getPos(); // buffer.tell(buffer.data); /* pointer to file and track headers */
|
||||
process_header();
|
||||
printf(" Processing %d tracks.\n", num_tracks);
|
||||
if (num_tracks > MAX_TRACKS) {
|
||||
midi_error("Too many tracks", buffer.tell(buffer.data));
|
||||
midi_error("Too many tracks", file->getPos()); //buffer.tell(buffer.data));
|
||||
}
|
||||
|
||||
/* initialize processing of all the tracks */
|
||||
@@ -454,7 +440,7 @@ int AudioGeneratorMIDI::PlayMIDI() {
|
||||
for (tgnum = 0; tgnum < num_tonegens; ++tgnum) { /* find which generator is playing it */
|
||||
tg = &tonegen[tgnum];
|
||||
if (tg->playing && tg->track == tracknum && tg->note == trk->note) {
|
||||
tsf_note_off_fast(g_tsf, tg->instrument, tg->note, tg->playIndex);
|
||||
tsf_note_off/*_fast*/(g_tsf, tg->instrument, tg->note);//, tg->playIndex);
|
||||
tg->playing = false;
|
||||
tg->playIndex = -1;
|
||||
trk->tonegens[tgnum] = false;
|
||||
@@ -489,7 +475,7 @@ int AudioGeneratorMIDI::PlayMIDI() {
|
||||
if (tg->instrument != midi_chan_instrument[trk->chan]) { /* new instrument for this generator */
|
||||
tg->instrument = midi_chan_instrument[trk->chan];
|
||||
}
|
||||
tg->playIndex = tsf_note_on_fast(g_tsf, tg->instrument, tg->note, trk->velocity / 127.0); // velocity = 0...127
|
||||
tg->playIndex = tsf_note_on/*_fast*/(g_tsf, tg->instrument, tg->note, trk->velocity / 127.0f); // velocity = 0...127
|
||||
} else {
|
||||
++notes_skipped;
|
||||
}
|
||||
@@ -502,15 +488,8 @@ int AudioGeneratorMIDI::PlayMIDI() {
|
||||
|
||||
void AudioGeneratorMIDI::StopMIDI() {
|
||||
|
||||
buffer.close(buffer.data);
|
||||
file->close();
|
||||
tsf_close(g_tsf);
|
||||
printf(" %s %d tone generators were used.\n",
|
||||
num_tonegens_used < num_tonegens ? "Only" : "All", num_tonegens_used);
|
||||
if (notes_skipped)
|
||||
printf
|
||||
(" %d notes were skipped because there weren't enough tone generators.\n", notes_skipped);
|
||||
|
||||
printf(" Done.\n");
|
||||
}
|
||||
|
||||
|
||||
@@ -524,18 +503,17 @@ bool AudioGeneratorMIDI::begin(AudioFileSource *src, AudioOutput *out) {
|
||||
}
|
||||
memset(midi_chan_instrument, 0, sizeof(midi_chan_instrument));
|
||||
|
||||
g_tsf = tsf_load(&afsSF2);
|
||||
if (!g_tsf) {
|
||||
return false;
|
||||
}
|
||||
tsf_set_output(g_tsf, TSF_MONO, freq, -10 /* dB gain -10 */);
|
||||
g_tsf = _tsf;
|
||||
tsf_set_output(g_tsf, TSF_STEREO_INTERLEAVED, freq, -10 /* dB gain -10 */);
|
||||
|
||||
if (!out->SetRate(freq)) {
|
||||
return false;
|
||||
}
|
||||
if (!out->SetChannels(1)) {
|
||||
|
||||
if (!out->SetChannels(2)) {
|
||||
return false;
|
||||
}
|
||||
|
||||
if (!out->begin()) {
|
||||
return false;
|
||||
}
|
||||
@@ -545,7 +523,7 @@ bool AudioGeneratorMIDI::begin(AudioFileSource *src, AudioOutput *out) {
|
||||
|
||||
running = true;
|
||||
|
||||
PrepareMIDI(src);
|
||||
PrepareMIDI();//src);
|
||||
|
||||
samplesToPlay = 0;
|
||||
numSamplesRendered = 0;
|
||||
@@ -557,40 +535,37 @@ bool AudioGeneratorMIDI::begin(AudioFileSource *src, AudioOutput *out) {
|
||||
|
||||
|
||||
bool AudioGeneratorMIDI::loop() {
|
||||
static int c = 0;
|
||||
|
||||
if (!running) {
|
||||
goto done; // Nothing to do here!
|
||||
file->loop();
|
||||
output->loop();
|
||||
return running;
|
||||
}
|
||||
|
||||
// First, try and push in the stored sample. If we can't, then punt and try later
|
||||
if (!output->ConsumeSample(lastSample)) {
|
||||
goto done; // Can't send, but no error detected
|
||||
}
|
||||
|
||||
// Try and stuff the buffer one sample at a time
|
||||
do {
|
||||
#ifdef ESP8266
|
||||
static int c = 0;
|
||||
c++;
|
||||
if (c % 44100 == 0) {
|
||||
if (c == 44100) {
|
||||
c = 0;
|
||||
yield();
|
||||
}
|
||||
|
||||
play:
|
||||
#endif
|
||||
|
||||
if (sentSamplesRendered < numSamplesRendered) {
|
||||
lastSample[AudioOutput::LEFTCHANNEL] = samplesRendered[sentSamplesRendered];
|
||||
lastSample[AudioOutput::RIGHTCHANNEL] = samplesRendered[sentSamplesRendered];
|
||||
sentSamplesRendered++;
|
||||
auto toSend = numSamplesRendered - sentSamplesRendered;
|
||||
auto ret = output->ConsumeSamples(samplesRendered + sentSamplesRendered * 2, toSend);
|
||||
sentSamplesRendered += ret;
|
||||
if (toSend != ret) {
|
||||
break;
|
||||
}
|
||||
} else if (samplesToPlay) {
|
||||
numSamplesRendered = sizeof(samplesRendered) / sizeof(samplesRendered[0]);
|
||||
if ((int)samplesToPlay < (int)(sizeof(samplesRendered) / sizeof(samplesRendered[0]))) {
|
||||
numSamplesRendered = (sizeof(samplesRendered) / sizeof(samplesRendered[0])) / 2;
|
||||
if ((int)samplesToPlay < (int)(sizeof(samplesRendered) / sizeof(samplesRendered[0])) / 2) {
|
||||
numSamplesRendered = samplesToPlay;
|
||||
}
|
||||
tsf_render_short_fast(g_tsf, samplesRendered, numSamplesRendered, 0);
|
||||
lastSample[AudioOutput::LEFTCHANNEL] = samplesRendered[0];
|
||||
lastSample[AudioOutput::RIGHTCHANNEL] = samplesRendered[0];
|
||||
sentSamplesRendered = 1;
|
||||
tsf_render_short(g_tsf, samplesRendered, numSamplesRendered, 0);
|
||||
samplesToPlay -= numSamplesRendered;
|
||||
sentSamplesRendered = 0;
|
||||
} else {
|
||||
numSamplesRendered = 0;
|
||||
sentSamplesRendered = 0;
|
||||
@@ -602,12 +577,10 @@ play:
|
||||
sawEOF = true;
|
||||
samplesToPlay = freq / 2;
|
||||
}
|
||||
goto play;
|
||||
}
|
||||
}
|
||||
} while (running && output->ConsumeSample(lastSample));
|
||||
} while (running);
|
||||
|
||||
done:
|
||||
file->loop();
|
||||
output->loop();
|
||||
|
||||
@@ -619,47 +592,3 @@ bool AudioGeneratorMIDI::stop() {
|
||||
output->stop();
|
||||
return true;
|
||||
}
|
||||
|
||||
|
||||
int AudioGeneratorMIDI::afs_read(void *data, void *ptr, unsigned int size) {
|
||||
AudioFileSource *s = reinterpret_cast<AudioFileSource *>(data);
|
||||
return s->read(ptr, size);
|
||||
}
|
||||
|
||||
int AudioGeneratorMIDI::afs_tell(void *data) {
|
||||
AudioFileSource *s = reinterpret_cast<AudioFileSource *>(data);
|
||||
return s->getPos();
|
||||
}
|
||||
|
||||
int AudioGeneratorMIDI::afs_skip(void *data, unsigned int count) {
|
||||
AudioFileSource *s = reinterpret_cast<AudioFileSource *>(data);
|
||||
return s->seek(count, SEEK_CUR);
|
||||
}
|
||||
|
||||
int AudioGeneratorMIDI::afs_seek(void *data, unsigned int pos) {
|
||||
AudioFileSource *s = reinterpret_cast<AudioFileSource *>(data);
|
||||
return s->seek(pos, SEEK_SET);
|
||||
}
|
||||
|
||||
int AudioGeneratorMIDI::afs_close(void *data) {
|
||||
AudioFileSource *s = reinterpret_cast<AudioFileSource *>(data);
|
||||
return s->close();
|
||||
}
|
||||
|
||||
int AudioGeneratorMIDI::afs_size(void *data) {
|
||||
AudioFileSource *s = reinterpret_cast<AudioFileSource *>(data);
|
||||
return s->getSize();
|
||||
}
|
||||
|
||||
void AudioGeneratorMIDI::MakeStreamFromAFS(AudioFileSource *src, tsf_stream *afs) {
|
||||
afs->data = reinterpret_cast<void*>(src);
|
||||
afs->read = &afs_read;
|
||||
afs->tell = &afs_tell;
|
||||
afs->skip = &afs_skip;
|
||||
afs->seek = &afs_seek;
|
||||
afs->close = &afs_close;
|
||||
afs->size = &afs_size;
|
||||
}
|
||||
|
||||
#endif //__GNUC__ == 8
|
||||
|
||||
+17
-18
@@ -21,22 +21,25 @@
|
||||
#ifndef _AUDIOGENERATORMIDI_H
|
||||
#define _AUDIOGENERATORMIDI_H
|
||||
|
||||
#if defined(ESP32) && (__GNUC__ >= 8) && (__XTENSA__)
|
||||
#if defined(xxxESP32) && (__GNUC__ >= 8) && (__XTENSA__)
|
||||
// Do not build, Espressif's GCC8+ has a compiler bug
|
||||
#else // __GNUC__ == 8
|
||||
|
||||
#include "AudioGenerator.h"
|
||||
|
||||
#define TSF_NO_STDIO
|
||||
#define TSF_CONST_FILE
|
||||
#define TSF_SAMPLES_SHORT
|
||||
#include "libtinysoundfont/tsf.h"
|
||||
|
||||
class AudioGeneratorMIDI : public AudioGenerator {
|
||||
public:
|
||||
AudioGeneratorMIDI() {
|
||||
freq = 44100;
|
||||
freq = 22050;
|
||||
running = false;
|
||||
};
|
||||
virtual ~AudioGeneratorMIDI() override {};
|
||||
#if 0
|
||||
bool SetSoundfont(AudioFileSource *newsf2) {
|
||||
if (isRunning()) {
|
||||
return false;
|
||||
@@ -45,6 +48,14 @@ public:
|
||||
MakeStreamFromAFS(sf2, &afsSF2);
|
||||
return true;
|
||||
}
|
||||
#endif
|
||||
bool SetSoundFont(tsf *t) {
|
||||
if (isRunning()) {
|
||||
return false;
|
||||
}
|
||||
_tsf = t;
|
||||
return true;
|
||||
}
|
||||
bool SetSampleRate(int newfreq) {
|
||||
if (isRunning()) {
|
||||
return false;
|
||||
@@ -62,10 +73,6 @@ public:
|
||||
private:
|
||||
int freq;
|
||||
tsf *g_tsf;
|
||||
struct tsf_stream buffer;
|
||||
struct tsf_stream afsMIDI;
|
||||
struct tsf_stream afsSF2;
|
||||
AudioFileSource *sf2;
|
||||
AudioFileSource *midi;
|
||||
|
||||
protected:
|
||||
@@ -138,7 +145,6 @@ protected:
|
||||
}; /* no more data left in this track */
|
||||
|
||||
|
||||
|
||||
/* portable string length */
|
||||
int strlength(const char *str) {
|
||||
int i;
|
||||
@@ -172,24 +178,17 @@ protected:
|
||||
|
||||
unsigned long get_varlen(int *ptr);
|
||||
void find_note(int tracknum);
|
||||
void PrepareMIDI(AudioFileSource *src);
|
||||
void PrepareMIDI();//AudioFileSource *src);
|
||||
int PlayMIDI();
|
||||
void StopMIDI();
|
||||
|
||||
// tsf_stream <-> AudioFileSource
|
||||
static int afs_read(void *data, void *ptr, unsigned int size);
|
||||
static int afs_tell(void *data);
|
||||
static int afs_skip(void *data, unsigned int count);
|
||||
static int afs_seek(void *data, unsigned int pos);
|
||||
static int afs_close(void *data);
|
||||
static int afs_size(void *data);
|
||||
void MakeStreamFromAFS(AudioFileSource *src, tsf_stream *afs);
|
||||
|
||||
int samplesToPlay;
|
||||
bool sawEOF;
|
||||
int numSamplesRendered;
|
||||
int sentSamplesRendered ;
|
||||
short samplesRendered[256];
|
||||
short samplesRendered[256 * 2];
|
||||
|
||||
tsf *_tsf = nullptr;
|
||||
};
|
||||
|
||||
#endif //__GNUC__ == 8
|
||||
|
||||
@@ -64,4 +64,3 @@ protected:
|
||||
};
|
||||
|
||||
#endif
|
||||
|
||||
@@ -69,7 +69,7 @@ bool AudioOutputPWM::begin() {
|
||||
}
|
||||
|
||||
bool AudioOutputPWM::ConsumeSample(int16_t sample[2]) {
|
||||
|
||||
//Serial.printf("%d\n", sample[0]);
|
||||
if (!pwmOn) {
|
||||
return false;
|
||||
}
|
||||
|
||||
File diff suppressed because it is too large.
Load diff
@@ -1,5 +1,4 @@
|
||||
Copyright (C) 2017 Bernhard Schelling
|
||||
Based on SFZero, Copyright (C) 2012 Steve Folta (https://github.com/stevefolta/SFZero)
|
||||
Copyright (C) 2017-2023 Bernhard Schelling (Based on SFZero, Copyright (C) 2012 Steve Folta, https://github.com/stevefolta/SFZero)
|
||||
|
||||
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
|
||||
|
||||
@@ -1,40 +0,0 @@
|
||||
This is a ported and significantly modified version of TinySoundFont by
|
||||
Bernhard Schelling. https://github.com/schellingb/TinySoundFont
|
||||
|
||||
TinySoundFont lets you use a SoundFont2 library (sampled MIDI instruments)
|
||||
to play high quality music with a single #include.
|
||||
|
||||
Porting and modifications done by Earle F. Philhower, III
|
||||
<earlephilhower@yahoo.com> and released under the GPL v3 or later.
|
||||
|
||||
The changes to allow it to work on the ESP8266 consisted mostly in making
|
||||
it significantly less memory intensive and speeding up the inner loops by
|
||||
using fixed point arithmetic instead of floating point. This change to
|
||||
fixed point lets us use simple integer math, which is massively faster than
|
||||
the software FP libraries on the ESP8266, but does come at the cost of a
|
||||
slightly increased noise floor.
|
||||
|
||||
On the memory side, all data structures were converted into a "lazy"
|
||||
allocation which means that instead of loading the full contents of the
|
||||
header bits into RAM (which is often WAY more than the 40KB available on
|
||||
the EPS8266), only those bits needed are loaded, and only when they are
|
||||
needed.
|
||||
|
||||
Also, instead of loading the entire sample array into memory (i.e. nearly
|
||||
500MB for a good piano sample), only a very small portion is loaded into a
|
||||
LRU *cache* area.
|
||||
|
||||
Even with the caching, it was found that SPIFFS, while having great
|
||||
functionality, was horrbly slow. So I wrote a new "faster" ROM filesystem
|
||||
called, surprisingly, FastROMFilesystem.
|
||||
https://github.com/earlephilhower/ESP8266FastROMFS
|
||||
If you are getting choppy playback, try this new filesystem or using a SD
|
||||
card (not tested by myself, but it'd be hard top be slower than SPIFFS).
|
||||
Simply going from SPIFFS to FastROMFilesystem took my testing of
|
||||
FURELISE.MID and 1MGM.SF2 from 0.5x realtime to 2.5x (i.e. from unusable
|
||||
stuttering to plenty of time to do other tings on the ESP8266) for the
|
||||
first 20 seconds or so tested.
|
||||
|
||||
-Earle F. Philhower, III
|
||||
earlephilhower@yahoo.com
|
||||
|
||||
@@ -0,0 +1,43 @@
|
||||
# Fixed-point MIDI ROM-wavetable synthesis
|
||||
|
||||
This branch was ported to ESP8266Audio by Earle F. Philhower, III <earlephilhower@yahoo.com>
|
||||
from the original TinySoundFont by https://github.com/schellingb/TinySoundFont
|
||||
|
||||
Changes include converting a large portion of the operation to fixed-point
|
||||
arithmetic and adding in a special ROM-mode where the SoundFont can be
|
||||
stored inside a microcontroller ROM w/o needing a filesystem. This saves
|
||||
RAM and makes the code faster than having to page in little bits of
|
||||
waveforms or regions.
|
||||
|
||||
## Building ROM-wavetable structures
|
||||
|
||||
The SoundFont is stored in ROM using the normal C structures TSF uses, so it
|
||||
requires a conversion from raw SF2 file into a header with structure members.
|
||||
Do NOT use `xxd` or similar on the SoundFont, it won't work.
|
||||
|
||||
In the `examples` directory, run `build-linux-gcc`to compile the SF2-to-H
|
||||
converted. Run it as follows:
|
||||
````
|
||||
$ ./dump-linux-x86_64 INPUT.SF2 OUTPUT.H
|
||||
````
|
||||
|
||||
For example
|
||||
````
|
||||
$ ./# dump-linux-x86_64 Scratch2010.sf2 scratch2010.h
|
||||
````
|
||||
|
||||
## Building the F16P16 tables
|
||||
|
||||
This port implements a few functions that normally require floating point
|
||||
math with piecewise linear approximations. They are hardcoded in the TSF.
|
||||
file already, but to regenerate them just run `make_f16p16_tables`.
|
||||
|
||||
## Performance
|
||||
|
||||
On devices that can perform hardware single-precision FP (i.e. the Pico 2,
|
||||
original ESP32, others) this code can generate up to 44.1khz sounds with
|
||||
over a dozen simultaneous voices without underflow and leaving some CPU
|
||||
for other work.
|
||||
|
||||
On devices without an FPU, sample rates of up to 22050hz are possible,
|
||||
depending on the MIDI and SF2 used.
|
||||
@@ -1,37 +1,37 @@
|
||||
# TinySoundFont
|
||||
SoundFont2 synthesizer library in a single C/C++ file
|
||||
|
||||
## Overview
|
||||
|
||||
TinySoundFont is a software synthesizer using SoundFont2 sound bank files.
|
||||
|
||||
The library is a single C header file so it is extremely simple to integrate in your C/C++ projects.
|
||||
|
||||
```c++
|
||||
#define TSF_IMPLEMENTATION
|
||||
#include "tsf.h"
|
||||
|
||||
...
|
||||
|
||||
tsf* TinySoundFont = tsf_load_filename("soundfont.sf2");
|
||||
tsf_set_output(TinySoundFont, TSF_MONO, 44100, 0); //sample rate
|
||||
tsf_note_on(TinySoundFont, 0, 60, 1.0f); //preset 0, middle C
|
||||
short HalfSecond[22050]; //synthesize 0.5 seconds
|
||||
tsf_render_short(TinySoundFont, HalfSecond, 22050, 0);
|
||||
```
|
||||
|
||||
The library code is based on [SFZero by Steve Folta](https://github.com/stevefolta/SFZero).
|
||||
|
||||
## Documentation
|
||||
|
||||
The API documentation can be found on [top of the library source code](https://github.com/schellingb/TinySoundFont/blob/master/tsf.h).
|
||||
|
||||
There are also [examples available](https://github.com/schellingb/TinySoundFont/tree/master/examples) which come with a sample SoundFont file and build and play sound on Win32, Win64, Linux and MacOSX with no further dependencies.
|
||||
|
||||
## Dependencies
|
||||
|
||||
C standard libraries for fopen, math and malloc (can be removed by providing custom functions with #defines).
|
||||
|
||||
## License
|
||||
|
||||
TinySoundFont is available under the [MIT license](https://choosealicense.com/licenses/mit/).
|
||||
# TinySoundFont
|
||||
SoundFont2 synthesizer library in a single C/C++ file
|
||||
|
||||
## Overview
|
||||
|
||||
TinySoundFont is a software synthesizer using SoundFont2 sound bank files.
|
||||
|
||||
The library is a single C header file so it is extremely simple to integrate in your C/C++ projects.
|
||||
|
||||
```c++
|
||||
#define TSF_IMPLEMENTATION
|
||||
#include "tsf.h"
|
||||
|
||||
...
|
||||
|
||||
tsf* TinySoundFont = tsf_load_filename("soundfont.sf2");
|
||||
tsf_set_output(TinySoundFont, TSF_MONO, 44100, 0); //sample rate
|
||||
tsf_note_on(TinySoundFont, 0, 60, 1.0f); //preset 0, middle C
|
||||
short HalfSecond[22050]; //synthesize 0.5 seconds
|
||||
tsf_render_short(TinySoundFont, HalfSecond, 22050, 0);
|
||||
```
|
||||
|
||||
The library code is based on [SFZero by Steve Folta](https://github.com/stevefolta/SFZero).
|
||||
|
||||
## Documentation
|
||||
|
||||
The API documentation can be found on [top of the library source code](https://github.com/schellingb/TinySoundFont/blob/master/tsf.h).
|
||||
|
||||
There are also [examples available](https://github.com/schellingb/TinySoundFont/tree/master/examples) which come with a sample SoundFont file and build and play sound on Win32, Win64, Linux and MacOSX with no further dependencies.
|
||||
|
||||
## Dependencies
|
||||
|
||||
C standard libraries for fopen, math and malloc (can be removed by providing custom functions with #defines).
|
||||
|
||||
## License
|
||||
|
||||
TinySoundFont is available under the [MIT license](https://choosealicense.com/licenses/mit/).
|
||||
File diff suppressed because it is too large.
Load diff
+2530
-2492
File diff suppressed because it is too large.
Load diff
Reference in new issue
Block a user