This file is no longer used. For detailed Changelog information, please refer to the
version control system's commits. For an overview of differences between versions,
For developer related "What's new"-information, doc/fluidsynth-v11-devdoc contains
=== OLD ===
2009-05-01 Pedro Lopez-Cabanillas <firstname.lastname@example.org>
* configure.ac: fix for win32 build.
2009-05-01 Pedro Lopez-Cabanillas <email@example.com>
* doc/Doxyfile: added fluid_filerenderer.c to Doxygen documentation.
* doc/fluidsynth-v11-devdoc.txt: license changed to CC-BY-SA 3.0
* doc/fluidsynth_arpeggio.c: new example added.
* doc/fluidsynth_metronome.c: new example added.
* include/fluidsynth.h, include/fluidsynth/audio.h,
include/fluidsynth/settings.h: Doxygen documentation.
* src/fluid_settings.c: Doxygen documentation.
2009-04-27 Josh Green <firstname.lastname@example.org>
* include/fluidsynth/audio.h: Moved new filerenderer documentation to source file.
* src/config_win32.h.in: Added 'typedef int socklen_t;' to the correct place.
* src/fluid_filerenderer.c: Removed 2 extra pasted duplicates of the file, moved
Doxygen documentation from header file and added "API 1.1.0" designators.
* src/config_win32.h: Removed from subversion, since it is generated from
2009-04-26 Josh Green <email@example.com>
* configure.ac: Added glib 2.10 as a dependency, added notes in output
for LASH, LADCCA and READLINE that they are GPL.
* src/fluid_io.c: Moved code to fluid_sys.c and removed.
* src/config_win32.h: Added "typedef int socklen_t" definition.
* src/fluid_defsfont.h: Removed glib ripped code.
* src/fluid_oss.c: Fixed warnings where return value of write() was
* src/fluid_sys.c: Re-organized, implemented portable fluid_curtime() and
fluid_utime() using glib functions and removed old platform specific
code, implemented fluid_thread functionality using glib and removed
old platform specific code, fluid_istream_readline(), fluid_istream_gets()
and fluid_ostream_printf() should now work on WIN32 also, added code
for WIN32 for TCP sockets (not yet tested).
* src/fluid_sys.h: Added fluid_gerror_message() macro to extract message
safely from GError structures, replaced fluid_mutex macros with
portable implementations using glib, removed new_fluid_client_socket()
and delete_fluid_client_socket() which were never implemented or used.
* src/fluidsynth.c: Added call to g_thread_init().
* src/fluidsynth_priv.h: Integer types now use glib integer types.
2009-04-11 Josh Green <firstname.lastname@example.org>
* FluidSynth release 1.0.9 "A Sound Future"
* configure.ac: Bumped version, no library interfaces added, removed or changed.
* doc/Makefile.am: Removed html and api folders from EXTRA_DIST.
* src/fluid_synth.c (fluid_synth_program_change): Preset substitute warning
now outputs MIDI channel.
2009-04-02 Pedro Lopez-Cabanillas <email@example.com>
* src/config_win32.h: fix compilation under MSVC 2008 and older
2009-03-15 Josh Green <firstname.lastname@example.org>
* ltconfig: Removed obsolete ltconfig script by suggestion of
* doc/fluidsynth.1: Some fixes from Sven Hoexter.
* src/fluid_adriver.c: Re-order of default drivers to jack, alsa, pulse.
* src/fluidsynth.c (fluid_synth_program_change): Added preset selection
fallback logic: Melodic - Fallback to Bank0:prognum followed by
Bank0:Program0, Percussion - Fallback to 128:0, code re-organization.
2009-03-08 Josh Green <email@example.com>
* src/fluid_jack.c: Added support for Jack MIDI.
* src/fluid_mdriver.c: Registered Jack MIDI driver.
* README-OSX: Update from Ebrahim Mayat.
2009-02-28 Pedro Lopez-Cabanillas <firstname.lastname@example.org>
* src/fluid_midi.c: Fix for ticket #22 (Wrong tempo changes)
* src/fluid_midi.h: delta-time accumulator moved to fluid_midi_file struct.
2009-02-03 Josh Green <email@example.com>
* Applied patch from KO Myung-Hun for OS/2 support including Dart audio
2009-01-29 Josh Green <firstname.lastname@example.org>
* src/Makefile.am: Added PortAudio driver conditional build.
* src/fluid_adriver.c: Registered fluid_portaudio_driver_settings.
* src/fluid_portaudio.c: Completely overhauled for Portaudio 19.
This driver appears to have been unbuildable before.
2009-01-08 Pedro Lopez-Cabanillas <email@example.com>
* configure.ac: detection of CoreMIDI support. Ticket #18.
* src/Makefile.am: conditional build of CoreMIDI driver.
* src/fluid_coremidi.c: Basic CoreMIDI driver.
* src/fluid_mdriver.c: added CoreMIDI driver.
2009-01-08 Josh Green <firstname.lastname@example.org>
* configure.ac: Followed GTK's lead for some unexplained magic for
stupid libtool version parameters (fixes autogen.sh bomb on
undefined macro LT_REVISION/LT_CURRENT/LT_AGE). Added
AC_CONFIG_MACRO_DIR([m4]) as suggested by libtoolize.
* Makefile.am: Added ACLOCAL_AMFLAGS=-I m4 as suggested by libtoolize.
2008-12-23 Josh Green <email@example.com>
* configure.ac: Added detection of PulseAudio driver.
* src/Makefile.am: Added conditional build of PulseAudio driver.
* src/fluid_adriver.c: Added PulseAudio driver and re-sorted drivers
by use preference.
* src/fluid_chan.c: Using MIDI enums for initializing channel CC values,
added supported for RPN GM MIDI messages Bend Range, Fine Tune and
Coarse Tune, added check for out of range NRPN parameters.
* src/fluid_midi.h: Added RPN enum midi_rpn_event.
* src/fluid_pulse.c: New PulseAudio driver.
2008-09-22 Pedro Lopez-Cabanillas <firstname.lastname@example.org>
* src/fluid_dsound.c: Fix for ticket #16 - dsound device can't be selected.
2008-09-07 Josh Green <email@example.com>
* src/fluid_alsa.c (new_fluid_alsa_seq_driver): Patch from
Nicolas Boulicault to add ALSA sequencer midi.portname setting.
* src/fluid_conv.h: S. Christian Collins' patch - changed
FLUID_ATTEN_POWER_FACTOR from -531.509 to -200.0.
* src/fluid_defsfont.c (fluid_defpreset_noteon): S. Christian Collins'
patch - crash bug fix related to using certain modulators in a preset.
* src/fluid_mdriver.c: Pedro Lopez-Cabanillas' patch which adds a
* src/fluid_mod.c: S. Christian Collins' patch - Stop forcing velocity
based filtering and a couple of calculation fixes to transform functions.
* src/fluid_synth.c: Nicolas Boulicault's patch to add midi.portname
(fluid_synth_program_change): added fix to properly search for a
* src/fluid_synth.h: Changed FLUID_NUM_PROGRAMS to 128 and set
DRUM_INST_BANK to 128.
* src/fluid_voice.c (fluid_voice_write): S. Christian Collins' patch -
force velocity envelope value to be that of the previous stage when
switching from decay to sustain and filter calculation now uses
synthesizer baud rate rather than fixed at 44100.
(fluid_voice_update_param): S. Christian Collins' patch - Use multiplier
for GEN_ATTENUATION to be compatible with EMU10K1 cards.
* src/fluid_winmidi.c: Pedro Lopez-Cabanillas' patch which adds a
* src/fluidsynth.c: Nicolas Boulicault's patch which adds midi.portname
setting. Pedro Lopez-Cabanillas' patch which breaks out of argument
processing loop for non getopt option argument handling when a non
option is encountered and not using Readline.
2007-11-17 Josh Green <firstname.lastname@example.org>
* FluidSynth release 1.0.8 "Its about funky time!"
* configure.ac: Bumped LT_REVISION and added call to AM_PROG_CC_C_O macro.
* Makefile: Updated fluidsynth.prj to fluidsynth.anjuta
* README-OSX: Update from Ebrahim Mayat for OS X Leopard
* acinclude.m4: Fixed embedded main function in AM_PATH_READLINE macro.
2007-11-11 Josh Green <email@example.com>
* configure.ac: Added --enable-trap-on-fpe and --enable-fpe-check to
assist with Floating Point Exception debugging.
* src/fluid_chorus.c: Reverted the rest of the chorus "Effect level clip"
patch, until something better is devised.
* src/fluid_synth.c: Added support for trapping on Floating Point
Exceptions on GLIBC systems, to aid developers in tracking down FPEs
with gdb, removed buffer alignment hacks since they are no longer
needed (not using SSE currently).
* src/fluid_sys.c (fluid_time_config): Added check for a CPU freq
calculation of 0.0, since this test is inadequate to begin with and
was coming up as 0.0 on my laptop, causing a FPE. Will replace with
real timer functions, in the future.
* src/fluid_voice.c: Removed zap_almost_zero macro as it was buggy and
had issues which went away when gcc optimization was turned off and in
the case of !WITH_FLOAT was using abs() which is integer based and
would cause FPEs.
(fluid_voice_write): Removed a memory alignment hack and moved a call
to fluid_fpe_check() to a better location.
(fluid_voice_effects): Replaced zap_almost_zero with a call to fabs(),
added fluid_fpe_check() call.
* src/fluidsynth_priv.h: Removed FLUID_ALIGN16BYTE hack, as it is no
2007-11-10 Josh Green <firstname.lastname@example.org>
* doc/fluidsynth.1: Updated man page with current command line options and other changes (minor).
* include/fluidsynth/synth.h: Reverted "Effect level clip" patch as it seems to cause chorus
count to have a much lessor effect.
2007-09-20 Josh Green <email@example.com>
* Doc updates to AUTHORS and latest README-OSX from Ebrahim Mayat.
* src/config_win32.h.in: VERSION is now filled in at configure time.
* src/fluid_alsa.c (fluid_alsa_audio_run_s16): Fixed bug which was
causing weird crashes with QSynth when new_fluid_audio_driver2() when
audio meters were enabled (user data parameter was being used as a
fluid_synth_t instance). Synth instance is now no longer used in
this case (it was only used for 16 bit dithering before).
* src/fluid_oss.c: Fixed the same bug that was affecting ALSA driver.
* src/fluid_rev.c: Reverted to old commented out code in regards to
* src/fluid_synth.c (fluid_synth_dither_s16): Now no longer uses
fluid_synth_t instance, but accepts a pointer to an integer instead
for keeping track of dithering buffer index (all that the synth instance
was being used for).
* src/fluid_synth.c (fluid_synth_one_block): Reverted patch which
performs assignment of chorus and reverb levels in synthesis loop,
until a better scheme is devised (unnecessary CPU consumption).
* Added Visual Studio .sln and .vcproj files and some minor source
changes to get FluidSynth to build with it.
* Back-converted Visual Studio project to VC++ 6 project for users
using that build platform (not tested).
2007-09-02 Josh Green <firstname.lastname@example.org>
* configure.ac: Removed SSE and longlong related switches (SSE support
removed for now and longlong is now always used).
* : Applied effect level clip patch from David Hilvert
* : Applied reverb damp scaling patch from David Hilvert
* src/fluid_dsp_float.c: No longer being #include'd and all interpolation
functionality has been re-written as separate functions, interpolating
around loops is now supported, effect (reverb/chorus/pan/filter) stuff
moved to fluid_voice.c.
* src/fluid_phase.h: 64 bit unsigned integers are now used for phase
index/fraction sample pointers, modified macros accordingly.
* src/fluid_voice.c: Removed SSE code, fluid_voice_init() renamed to
fluid_dsp_float_init() and moved to fluid_dsp_float.c. Effect related
functionality (reverb/chorus/pan/filter) moved from fluid_dsp_float.c
to fluid_voice.c. Some code re-formatting and comment cleanup. Loop
no longer requires padding surrounding it (fixes bug related to loop
point right on the end of the sample).
2007-08-18 Josh Green <email@example.com>
* src/fluid_alsa.c: Added SND_SEQ_PORT_TYPE_MIDI_GENERIC back into the
ALSA sequencer port registration as it broke the use of playmidi
(thanks to Dave Serls for providing a patch and pointing this out).
2007-01-14 Josh Green <firstname.lastname@example.org>
* src/fluid_alsa.c: Fixed evil bugs in ALSA driver where return value
of new fluid_alsa_handle_write_error() was not being checked correctly
causing successfully handled ALSA errors (underruns for example) to
terminate audio thread.
* src/fluid_synth.c: Using an inline roundi function to replace roundf
as per suggestion by Mihail Zenkov, 16 bit for dithering.
2006-12-10 Josh Green <email@example.com>
Lots of documentation updates.
* doc/Doxyfile: No longer including functions by default, only those
listed in the listed header files.
* src/fluid_strtok.[ch]: Removed, since it was crap. Replaced with
fluid_strtok() in fluid_sys.c which doesn't require an allocated
* src/fluid_alsa.c: Audio processing is more optimized in the case where
no user defined audio callback is used (removal of unneeded buffer copy),
fluid_alsa_handle_write_error() added for centralized ALSA audio error
* src/fluid_aufile.c: Now also doing 16 bit dithering.
* src/fluid_cmd.c: Removed use of old tokenizer instance.
* src/fluid_coreaudio.c: User defined callback function is now honored.
* src/fluid_defsfont.c: More leaks plugged (thanks to Paul Millar for
the patch), removed sfont_free_data() since sfont_close() should be
used instead (don't want to leak a file handle).
* src/fluid_midi_router.c: Took out uses of fflush() since sending a
line of text (with newline) should display it.
* src/fluid_oss.c: Using fluid_synth_dither_s16() in place of old
16 bit conversion code.
* src/fluid_settings.c: Replaced strtok stuff with new function, some
* src/fluid_synth.c (delete_fluid_synth): Turning off all voices so that
SoundFont data will be freed correctly (thanks to patch from
* src/fluid_sys.c (fluid_strtok): New function to replace old tokenizing
functions which required a token instance.
* src/fluidsynth.c: Warning message printed if a non option is not a
valid SoundFont or MIDI file (thanks to Nick Daly for the patch).
2006-11-22 Josh Green <firstname.lastname@example.org>
* src/fluid_alsa.c (new_fluid_alsa_audio_driver2): Removed some ALSA
lib calls to set software parameters, which was likely causing the
100% CPU usage problem (not actually fixed in last update, not sure
which one is the culprit).
(fluid_alsa_seq_run): More changes in ALSA sequencer code, hopefully
it is right this time!
(delete_fluid_alsa_seq_driver): Memory leak fixed - wasn't freeing
array of sequencer file descriptors.
* src/fluid_chan.c: Memory leak fixes: Now deleting preset from channel
when channel is destroyed.
* src/fluid_cmd.c: Memory leak fix: strtok being deleted from command
shell when shell is destroyed.
* src/fluid_defsfont.c: Memory leak fixes: Freeing modulator lists in
preset and instrument zones, freeing zone names, freeing instruments
linked from preset zones, replaced use of "safe_malloc" with FLUID_MALLOC
macro, deleting instrument list in SFData, deleting samples in SFData,
freeing SFData structure.
* src/fluid_settings.c: Memory leak fix: freeing options in option
* src/fluid_synth.c: Memory leak fixes: Freeing FX buffers and
2006-11-21 Josh Green <email@example.com>
* src/fluid_alsa.c (new_fluid_alsa_audio_driver2): Modified all ALSA
calls to check return code error as "< 0" as per ALSA examples, sample
rate is now compared with what was expected and warning message displays
both values, if target sample rate wasn't set update the local
period_size variable (was causing 100% CPU consumption by ALSA, from
the resultant erroneous sw_params calls).
(fluid_alsa_audio_run_float): Using case statement for
error codes from snd_pcm_writen() for the sake of tidiness.
(fluid_alsa_audio_run_s16): Using case statement for error codes
from snd_pcm_writei() for the sake of tidiness, re-instated call
of device callback function that was broken with the dither patch
(don't want to break the API), now using new fluid_synth_dither_s16()
to convert floating point sample data to 16 bit with dithering.
(fluid_alsa_seq_run): Timeout in poll() call set to 100ms (from 1ms!),
snd_seq_event_input_pending is used to check if events are available
before calling snd_seq_event_input to prevent blocking, check of
snd_seq_event_input error code moved to the right location (bug fix).
* src/fluid_synth.h: Added dither_index parameter to fluid_synth_t
structure to allow for per synth dithering continuity.
* src/fluid_synth.c: Modified dithering to use new dither_index field
for per synth dithering continuity, fixed off by 1 error with
dithering index comparison, removed usage of roundf in dithering (is
it sufficient to just integer truncate?).
(fluid_synth_dither_s16): New function to perform dithering on
buffers of floating point sample data.
2006-11-20 Josh Green <firstname.lastname@example.org>
* src/fluid_alsa.c: Applied dithering patch from Mihail Zenkov.
* src/fluid_synth.c: Applied dithering patch from Mihail Zenkov.
2006-03-04 Josh Green <email@example.com>
* src/fluid_alsa.c (delete_fluid_alsa_audio_driver): Now calling
snd_pcm_close() to close the ALSA audio driver handle.
(fluid_alsa_seq_run): Check for -ENOSPC error was logicly inverted.
(new_fluid_alsa_seq_driver): Sequencer is now opened in blocking mode.
2006-02-20 Josh Green <firstname.lastname@example.org>
* Fixed build error that occured when neither LASH or LADCCA are
* Updated README-OSX from Ebrahim Mayat.
2006-02-18 Josh Green <email@example.com>
* FluidSynth release 1.0.7 "Increasing Fluidity.."
* Removed spurious newlines from FLUID_LOG statements throughout.
* AUTHORS: Some cleanup and additions.
* src/fluid_lash.[ch]: Moved LADCCA related code from fluidsynth.c here
and added new LASH support (both old LADCCA and LASH are supported
exclusively). Used patches sent by Frieder Bürzele as a guide.
* src/fluidsynth.c: Removed LADCCA code (now in fluid_lash.c),
re-organized command line parsing and removed duplicate WIN32 switch
statement, re-organized help output and added missing entries, added
"-o help" switch for listing settings, welcome message now printed
whenever FluidSynth is run and simplified,
(print_usage): hard coded application name as "fluidsynth".
* configure.ac: Changed --enable-SSE option to --enable-broken-SSE
and --enable-SSE now just displays a fat warning about not using it.
* src/fluid_jack.c: Warning is now displayed if synth sample rate
doesn't match jackd.
* src/fluid_alsa.c: Added detection for ALSA sequencer buffer overrun
(-ENOSPC) and interrupted poll() call (-1??).
* src/fluid_voice.c: Applied patch from Henri Manson which adds a
fluid_ct2hz_real() function which does not have the filter cutoff
limits that fluid_ct2hz() does, new function being used for
calculations that may include non-audible frequencies.
* src/fluid_dsound.c: Applied patch from Henri Manson which
only creates the directsound window once.
2005-09-04 Josh Green <firstname.lastname@example.org>
* src/fluid_ramsfont.c (fluid_ramsfont_remove_izone): Applied crash bug
fix from Antoine Schmitt.
2005-07-05 Josh Green <email@example.com>
* src/fluidsynth_priv.h: FLUID_ALIGN16BYTE is broken on AMD64 so now
only enabled if SSE is being used. If SSE code becomes more useful in
the future this should be fixed.
2005-06-29 Josh Green <firstname.lastname@example.org>
* Applied LASH patch that is included with ladcca-0.4.0.
* Released FluidSynth 1.0.6 "Music to my ears"
* README-OSX: Update from Ebrahim Mayat.
* acinclude.m4: Midishare support now defaults to auto.
* configure.ac: Added LT_CURRENT, LT_REVISION and LT_AGE in place of
LIBFLUIDSYNTH_MAJ and LIBFLUIDSYNTH_MIN to make better use of
libtool library versioning. Fixed use of AC_ARG_ENABLE (was setting
variables to yes even when disable was specified), fixes --disable-SSE
which was reported by Mikhail Yakshin, added warning when SSE is
enabled to let users know that this feature isn't really desirable
* src/Makefile.am: Now using LT_VERSION_INFO to substitute the libtool
* src/fluid_cmd.c (fluid_cmd_handler_handle): Modified to avoid GCC
"type-punned" cast warning.
* src/fluid_defsfont.c (fluid_preset_zone_import_sfont): Fixed
assignment of modulator amtsrc flags (should be assigned to flags2
not flags1), thanks to Stephan Tassart for reporting this.
(fluid_inst_zone_import_sfont): Same fixes as for above.
* src/fluid_sys.c (fluid_log): Now using vsnprintf for formatting
error messages to fix buffer overflow as reported by Axioplase.
(fluid_debug): Same as above.
* fluidsynth.prj: Added Anjuta project file.
* src/fluid_conv.c: fluid_cb2amp conversion set back to real centibels
and added a new fluid_atten2amp table conversion for non-standard
EMU 8k/10k attenuation.
* src/fluid_voice.c (fluid_voice_write): Updated volume calculations to
use fluid_cb2amp for envelope and LFO, but use fluid_atten2amp for
(fluid_voice_noteoff): Re-coded volenv_val attack conversion and
* src/fluid_phase.h: Patch from Sean Bolton to fix big endian long long
phase combined 64 bit value type fluid_phase_t
* src/fluid_voice.c (fluid_voice_update_param): case GEN_OVERRIDEROOTKEY
was incorrectly adding pitchadj fine tune amount instead of subtracting
it. Also, fine tuning should be applied to root key override as well.
* Applied Sean Bolton's DSSI patch (SB patch) which adds the ability to
change polyphony at runtime and fixes a bug (see below).
* README-OSX: Update from Ebrahim Mayat for OSX Panther.
* include/fluidsynth/synth.h: Sean Bolton's DSSI patch adds two new
functions fluid_synth_set_polyphony and fluid_synth_get_polyphony.
* src/fluid_conv.c: Centibel to amplitude conversion now follows
EMU 8k/10k which is contrary to SoundFont specification (TiMidity++
used as an example).
* src/fluid_conv.h: FLUID_CB_POWER_FACTOR defined for the centibel->amp
conversion table equation.
* src/fluid_defsfont.c (load_pgen): Fixed 'use of cast expressions as
lvalues is deprecated' warning by casting the value being assigned
instead of the variable assigned to and removed code warrior
specific code to work around this.
(load_igen): Same as for load_pgen.
* src/fluid_synth.c: SB patch - uses synth->polyphony instead of
synth->nvoice when iterating over the synth's voices.
(fluid_synth_update_polyphony): SB patch (new) - runtime update
(fluid_synth_set_polyphony): SB patch (new)
(fluid_synth_get_polyphony): SB patch (new)
(fluid_synth_nwrite_float): SB patch - fixes bug where the use of
arbitrary values of the 'len' parameter was broken.
* src/fluid_voice.c (fluid_voice_write): modlfo_to_vol (modulation LFO
to volume) was being calculated inverted (should be negative
attenuation, gain, for a positive rise in LFO).
(fluid_voice_noteoff): Updated centibel to amplitude conversion
used when voice off during attack to use the new
* README-OSX: Update from Ebrahim Mayat.
* src/fluid_synth.c (fluid_synth_set_bank_offset):
(fluid_synth_get_bank_offset): New API to set
a bank offset in a SoundFont (proposition made by Ken Ellinwood).
* src/fluid_synth.c (fluid_synth_noteon):
fluid_synth_release_voice_on_same_note() is now called in the
noteon() function instead of in fluid_synth_start(). This fixes
the silent note problem!
* src/fluid_chan.c (fluid_channel_cc): Applied Ken Ellinwood's
fix for the bank select (MSB) message.
* src/fluid_jack.c (fluid_jack_audio_driver_settings): Applied Rui
Nuno Capela's patch
* doc/fluidsynth.1 (option): Fixed typo noted by Gerald Pye.
2004-05-14 Peter Hanappe <email@example.com>
* src/fluid_dsound.c (fluid_dsound_enum_callback): Applied Sergey
Pavlishin's patch. This path fix stack overflow during DirectSound
audio driver initialization.
2004-05-07 Peter Hanappe <firstname.lastname@example.org>
* src/fluid_synth.c (fluid_synth_remove_sfont): Added new function
2004-05-05 Peter Hanappe <email@example.com>
* src/fluid_alsa.c (new_fluid_alsa_seq_driver): The alsa driver
now opens several ports if the synthesizer is configured for more
than 16 MIDI channels.
* src/fluid_voice.c (fluid_voice_write): I removed the filter
on/off optimization. The filter is always on and serves as an
2004-05-04 Peter Hanappe <firstname.lastname@example.org>
* src/fluid_synth.c (new_fluid_synth): The number of MIDI channels
now has to be a multiple of 16. The synth checks that this is the
case and changes the settings accordingly. I removed the sanity
checks for the min/max value of the number of MIDI channels since
this is already done by the settings object.
2004-03-30 Josh Green <email@example.com>
* src/fluid_voice.c (fluid_voice_write): Altered filter turn-off
optimization to not turn filter off once it has been
enabled. There is still a potential for a click when it gets
turned on though, which needs to be dealt with.
2004-03-30 Peter Hanappe <firstname.lastname@example.org>
* src/fluid_dsp_core.c: I've split up the dsp core file in three
files: fluid_dsp_simple.c, fluid_dsp_float.c, and
fluid_dsp_sse.c. This improves the readability.
2004-03-29 Peter Hanappe <email@example.com>
* src/fluid_jack.c (new_fluid_jack_audio_driver2): Testing the
number of ports before allocating them.
the "audio.jack.autoconnect" setting.
* src/fluid_midi.c (fluid_player_set_midi_tempo): Tempo changes
handled correctly. Was broken after fix on [2004-03-22] (see
* src/fluid_strtok.c (fluid_strtok_char_index): Removed printf's
2004-03-26 Peter Hanappe <firstname.lastname@example.org>
* bindings/README: Imported the fluidsynth_jni and fluidmax
2004-03-25 Peter Hanappe <email@example.com>
* src/fluid_rev.c (new_fluid_revmodel): Added 'gain', similar as
in Freeverb 3. Using same 'wetscale' as Freeverb 3, but fixing
'wet' to 3. fluid_revmodel_setlevel() does not change the value of
'wet': The 'wet' level can be controlled with the 'reverb send'.
(fluid_revmodel_processreplace): The input is multiplied by 2 and
by the gain. This corresponds to the channel mixing and scaling
that Freeverb 3 does.
2004-03-24 Peter Hanappe <firstname.lastname@example.org>
* src/fluidsynth.c (main): Added the -f switch. Passing "-f file"
on the command line tells fluidsynth to read parse the file and
execute and commands.
(main): User config and system config file are now loaded correctly
* src/fluid_cmd.c (fluid_shell_run): the shell doesn't get stuck
and loop on an emtpy string when the end of the stream is reached.
* src/fluid_io.c (fluid_istream_gets): fluid_istream_gets()
returns 0 if the end of the stream was reached and -1 on error.
* src/fluid_cmd.c (fluid_source): Fixed bug in "file =
open(filename, FLAGS);" (I shouldn't pass O_WRONLY when what I
want is O_RDONLY!)
2004-03-23 Peter Hanappe <email@example.com>
* src/fluid_aufile.c (new_fluid_file_audio_driver): Added
fluid_aufile.c. This file implements a audio driver that writes
the audio output to a file. This driver is NOT real-time and is
currently useful for testing purposes only (not even useful to
play MIDI files).
2004-03-22 Peter Hanappe <firstname.lastname@example.org>
* src/fluid_synth.c (new_fluid_synth): Removed the synth->busy
mutex. I don't think it is necessary; to be discussed.
* src/fluid_midi.c (fluid_player_callback): Fixed the timing in
the MIDI playback. The current MIDI tick in every timer callback
was calculated as an increment to the previous number of
ticks. This introduces a growing error due to rounding errors and
timer variations. The current tick is now calculated according to
the absolute time at the beginning of the file. (Beginners error
* doc/FluidSynth-LADSPA.pdf: Added Markus' LADSPA design document.
* doc/xtrafluid.txt: Added Antoine's Xtra API documentation.
* doc/midi_time.txt: Added a memo on midi timing.
2004-03-19 Peter Hanappe <email@example.com>
* src/fluid_midishare.c: Applied Stephane Letz patch: MidiShare is
now connected to fluidsynth by default so that received MIDI
events directly trigger the synth
2004-02-28 Peter Hanappe <firstname.lastname@example.org>
* src/fluid_synth.c: Added fluid_synth_program_select2() and
fluid_synth_get_sfont_by_name() in fluid_synth.c. These functions
are not in the public API, yet.
2004-02-25 Peter Hanappe <email@example.com>
* src/fluid_voice.c: Fixed bug in volume envelope (in
fluid_voice_update_param(), case GEN_VOLENVDECAY): the minimum
value was converted to linear amplitude instead of a normalized
value of the cB (1-cB/1000). Because of that, the decay section
went on for too long.
2004-12-xx Peter Hanappe <firstname.lastname@example.org>
* src/fluid_seq.c: Inserting events in the queueLater list was
incomplete. It didn't check if the event was the last in the list,
and the looping through the list didn't update the prev pointer. I
added muteces to the sequencer. Events are dynamically allocated
if no free events are available. The sequencer is protected by a
2003-11-14 Josh Green <email@example.com>
* src/fluidsynth.c: Removed CCA_Use_Jack and CCA_Use_Alsa flags
since LADCCA no longer uses them.
2003-08-31 Josh Green <firstname.lastname@example.org>
* acinclude.m4: Renamed AC_SOUND macro to AC_OSS_AUDIO and removed
the ALSA check from it since pkg-config is now being used to check
for ALSA. Also fixed --enable-alsa-support and --enable-oss-support
which were disabling support instead (reported by Bart Massey).
* configure.ac: pkg-config is now being used to check for ALSA.
ALSA and OSS now use automake conditionals to conditionally compile
* Makefile.am: Re-arranged SUBDIRS so build output looks nicer.
* src/Makefile.am: ALSA and OSS are now conditionally compiled using
2003-08-29 Josh Green <email@example.com>
* src/fluid_sys.c: Patch from Eric Van Buggenhaut to make i386 asm
code not compile for all non-i386 archs rather than just DARWIN.
* src/fluidsynth_priv.h: Patch from Sergey Pavlishin to fix
* src/fluid_cmd.c: Ken Ellinwood's patch to add -verbose to "channels"
command, and print settings values with 3 decimal places.
* src/fluid_defsfont.c (fluid_defsfont_sfont_get_preset): Ken
Ellinwood's patch to initialize sfont field of preset.
* src/fluid_ramsfont.c (fluid_ramsfont_sfont_get_preset): Ken
Ellinwood's patch to initialize sfont field of preset.
* src/fluid_midi.c (fluid_midi_file_read_event): Fixed a crash bug with
zero length MIDI meta events that was pointed out by Sergey Pavlishin.
(delete_fluid_midi_event): Fixed a stack overflow problem pointed out
by Sergey Pavlishin that was caused by recursively deleting MIDI
event linked list, now just using a while loop.
2003-08-25 Josh Green <firstname.lastname@example.org>
* src/fluidsynth.c: MIDI channels switch should be -K not -L as was
listed in "Usage" output, also -K was setting audio.channels for
non getopt case statement - changed to midi.channels. Added a new
option "-l, --disable-ladcca" to disable LADCCA server connection.
2003-08-25 Josh Green <email@example.com>
Release version 1.0.3
* doc/fluidsynth.1: Applied typo patch from Eric Van Buggenhaut.
* TODO: Restructuring TODO file (removing old stuff).
* doc/Doxyfile: Disabled Tex doxygen generation and changed
OUTPUT_DIRECTORY to api/.
* doc/Makefile.am: Added an update-docs target and related for updating
developer doc and doxygen reference HTML. Also added update-docs to
dist-hook for updating before distribution packaging.
* include/fluidsynth/synth.h: Some fixes to doxygen documentation.
* fluidsynth.spec.in: New RPM spec file which is generated at configure
* Makefile.am: Added fluidsynth.spec(.in) to EXTRA_DIST.
2003-08-19 Josh Green <firstname.lastname@example.org>
* src/fluid_alsa.c: Added some calls to snd_strerror() to print out
details of ALSA routine failures.
* src/fluid_defsfont.c: Put a message about SoundFont loading code
being borrowed from Smurf SoundFont Editor.
* src/fluid_rev.c: Valgrind found that some values were being used
uninitialized because fluid_revmodel_update() was being called before
all reverb parameters were set, now setting manually and then calling
* src/fluid_voice.c: Increased FLUID_MAX_AUDIBLE_FILTER_FC to minimize
clicks from filter toggling. Added a FLUID_MIN_VOLENVRELEASE constant
to set the minimum volume envelope release to minimize clicks.
2003-07-22 Josh Green <email@example.com>
* src/fluid_midishare.c: Added include of header "config.h" as
per Albert Graef's request.
* src/fluid_voice.c (fluid_voice_optimize_sample): Moved a
variable declaration to the beginning of function, it was
causing problems with at least one user.
2003-06-28 Josh Green <firstname.lastname@example.org>
* src/fluid_defsfont.c: Moved call to fluid_voice_optimize_sample
from fluid_inst_zone_import_sfont to fluid_defsfont_load. Also
reduced minimum sample size before rejection from 48 to 8 (could
* src/fluid_voice.c (fluid_voice_optimize_sample): Added a check
for sample->valid to ignore ROM samples which was causing a crash
with Vintage Dreams and other SoundFont files with ROM samples.
2003-06-17 Josh Green <email@example.com>
Release version 1.0.2
Added Makefile.am files where lacking.
* Makefile.am: Fixes to "make dist" target by adding macbuild, sf2
and winbuild to SUBDIRS also removed acconfig.h from EXTRA_DIST.
* acinclude.m4: Removed AC_JACK, now using pkgconfig.
* configure.ac: Updated to version 1.0.2, Jack test now using
pkgconfig and built by default if found, coreaudio driver now built
by default if found.
* doc/Makefile.am: Added Doxyfile, example.c, example.sf2, fluidsynth.1
and fluidsynth-v10-devdoc.xml to EXTRA_DIST.
* src/Makefile.am: fluid_jack.c now conditionally built, fluid_sse.h
added to EXTRA_DIST.
* src/fluid_jack.c: #if JACK_SUPPORT removed as its not needed.
2003-06-15 Josh Green <firstname.lastname@example.org>
* configure.ac: Fixed detection of CoreAudio by looking for
* src/Makefile.am: Added COREAUDIO_CFLAGS and COREAUDIO_LIBS.
* src/fluid_coreaudio.c: Added CoreAudio prefix to #include headers
(fluid_core_audio_callback): Fixed declarition to match that of the
typedef in CoreAudio header to stop warnings.
* fluidsynth.c: Now including fluidsynth_priv.h to include the arch
specific definitions in there (perhaps should be done in configure
* fluidsynth_priv.h: Added "#define WITHOUT_SERVER 1" to Darwin build.
2003-06-12 Josh Green <email@example.com>
* Makefile.am: Added autogen.sh to EXTRA_DIST
* acinclude.m4: Added AM_PATH_READLINE macro for readline detection
and prefix configuration.
* configure.ac: Support for MinGW32 build, Darwin build fixes,
configure CFLAGS input value now honored, fixes to CoreAudio support,
and better readline detection and config.
* src/Makefile.am: Now conditionally compiling CoreAudio and Windows
sources, added config_*.h files to EXTRA_DIST, some stuff for MinGW32
build, READLINE_LIBS and READLINE_CFLAGS now used.
* src/fluid_dsound.c: Fixed some warnings by adding "void" for
empty parameter procedure declarations.
* src/fluidsynth.c: Don't include config_win32.h if MinGW32.
* src/fluidsynth_priv.h: Stuff for MinGW32 and Darwin builds.
* doc/fluidsynth-v10-devdoc.xml: Applied a diff from
2003-06-09 Josh Green <firstname.lastname@example.org>
* src/fluid_alsa.c: Added calls to pthread_attr_setschedparam to
properly create SCHED_FIFO threads.
* src/fluid_oss.c: pthread_attr_setschedparam calls added.
* src/fluid_midishare.c: Patch update from Stephane Letz.
2003-05-29 root <mn@bongo>
* src/fluid_synth.c (fluid_synth_one_block): Added a mutex that
provides a small degree of protection against noteons / noteoffs,
when the audio thread is working.
* src/fluid_synth.h (struct _fluid_synth_t):
* src/fluid_voice.c (fluid_voice_optimize_sample):
2003-05-29 Markus Nentwig <email@example.com>
* include/fluidsynth/voice.h: added fluid_voice_gen_incr to api
* src/fluidsynth.c: Added error message for command line parameter handling
* src/fluid_voice.c (fluid_voice_optimize_sample): Removed loop peak detection
at run time, because it caused dropouts. Now the sound font loader or application
is responsible to call fluid_voice_optimize_sample (if it doesn't, the turnoff optimization is
1999-11-30 Antoine Schmitt <firstname.lastname@example.org>
* src/fluid_defsfont.c: inst_zone lokey is now properly inialized to 0
(it was not, leading to random lost noteons depending on memory
2003-04-03 Peter Hanappe <email@example.com>
* src/fluid_rev.c: reverb parameters are clipped to their valid
* src/fluid_alsa.c: using fluid_alsa_audio_run_s16 as default
function. This reduces the high CPU usage.
* src/fluid_voice.c (fluid_voice_write): filter interpolation done
over only 1 buffer to avoid filter instability
* src/fluid_chan.c (fluid_channel_init): bank number set to 128
for the drum channel
* src/fluid_midi.c (fluid_midi_file_read_event): Correctly reading
2003-02-27 Josh Green <firstname.lastname@example.org>
Updated automake files (automake 1.6).
* configure.ac: New version autoconf variables which get substituted
* include/iiwusynth/version.h.in: Version defines that are filled in
* src/Makefile.am: Fixed SOURCES including removing headers that are
now in include/iiwusynth/, added missing sources (iiwu_ramsfont.[ch],
iiwu_sfont.h) and added iiwu_dsp_core.c to EXTRA_DIST.
* doc/Makefile.am: Added iiwusynth.1 to EXTRA_DIST.
* include/iiwusynth.h: Added version.h.
* iiwusynth/Makefile.am: Added version.h to the installed headers.
2003-02-08 Markus Nentwig <email@example.com>
* src/iiwu_ladspa.c: Added a very small signal at Nyquist
frequency. This fixes denormal number problems in some
* src/iiwu_cmd.c (iiwu_shell_run): Now also invalid input lines
are added to the command line history. So the user can just scroll
up and fix them.
* src/iiwu_ladspa.c: Cleaned up error messages
* src/iiwu_dsp_core.c: Disabled SSE interpolation, because it is
slower than the normal code
* autogen.sh: Added a line, that checks for the presence of pkg-config in
Motivation: It took me some time to figure out what was wrong...
It produces some error message instead of an obscure error later
during ./configure, if pkg-config is not installed.
2003-02-07 Josh Green <firstname.lastname@example.org>
Applied another Bob Ham LADCCA patch.
* src/iiwu_alsa.c: LADCCA patch: Now using a ladcca.enable setting.
* src/iiwu_jack.c: LADCCA patch: ladcca.enable setting and jack ports
are no longer auto connected unless audio.jack.autoconnect is set.
* src/iiwusynth.c: LADCCA patch: ladcca.enable and command line options
-j and --connect-jack-outputs to enable Jack autoconnect.
2003-02-05 Josh Green <email@example.com>
Applied Bob Ham's LADCCA and pkgconfig patches.
* Makefile.am: pkgconfig patch.
* configure.ac: Renamed from configure.in as per new autoconf
standards. LADCCA configure switch and detection. FluidSynth.pc
pkgconfig file output.
* src/Makefile.am: LADCCA patch.
* src/iiwu_alsa.c [HAVE_LADCCA]: LADCCA patch: reports ALSA
sequencer client ID.
* src/iiwu_jack.c [HAVE_LADCCA]: LADCCA patch: reports JACK
* src/iiwusynth.c [HAVE_LADCCA]: LADCCA patch: connects to LADCCA
server, creates client thread, saves/restores SoundFont file state.
Used iiwu_sfont_get_name macro to get SoundFont file names contrary
to the patch. Should these macros be public? Included unistd.h
for usleep call (within HAVE_LADCCA).
2003-01-23 Josh Green <firstname.lastname@example.org>
* src/iiwu_jack.c: Fixed a segfault bug caused by freeing jack
port names, when really only the port array should be freed,
jack reference docs are confusing on this matter!
* src/iiwu_voice.c (iiwu_voice_check_sample_sanity): Min loop size
and padding now set via constants IIWU_MIN_LOOP_SIZE and
IIWU_MIN_LOOP_PAD defined at top of iiwu_voice.c, and the values
were lowered to exceed SF spec requirements rather then just meet.
(iiwu_voice_write): Now using a constant IIWU_MAX_AUDIBLE_FILTER_FC
defined at the top of iiwu_voice.c to control the filter
cutoff optimization. Also added IIWU_MIN_AUDIBLE_FILTER_Q so
filter will only turn off if both cutoff and q are determined to be
inaudible. Filter optimization is much less noticeable when modulating.
2003-01-14 Markus Nentwig <email@example.com>
* src/iiwu_ladspa.c: Adapted new command handler
* src/iiwu_midi_router.c (midi_dump_prerouter): Added forgotten
'flush' for event dump 'fprintf's
2003-01-01 Markus Nentwig <firstname.lastname@example.org>
* src/iiwu_oss.c (new_iiwu_oss_audio_driver): Changed to callback function
* src/iiwu_alsa.c (new_iiwu_alsa_midi_driver): Changed to callback function
* src/iiwu_midishare.c (iiwu_midishare_midi_driver_receive): Partly done the same
* src/iiwu_winmidi.c (new_iiwu_winmidi_driver): To be done...
* src/iiwu_midi_router.c: Added
* src/iiwu_ladspa.c: Adapted to new settings system
* src/iiwu_adriver.c (iiwu_audio_driver_settings): Uses getint
instead of getnum for audio.period-size and audio.periods settings.
* src/iiwu_voice.c (iiwu_voice_write):
Ignore the valid flag for samples. Otherwise no sound is produced.
* src/iiwu_chan.c (iiwu_channel_cc): Fixed bank select (7-bit
instead of 8 bit)
2002-12-23 Peter Hanappe <email@example.com>
* src/iiwu_io.c (iiwu_istream_readline): new file (iiwu_io.c and
iiwu_io.c) to handle IO in the shell.
* src/iiwusynth.c (main): options to start TCP server.
* src/iiwu_cmd.c (new_iiwu_server): New structure and functions
(new_iiwu_shell): New structure and functions to improve command
(new_iiwu_cmd_handler): New structure and functions to improve command
* src/iiwu_sys.c (new_iiwu_server_socket): New structure and
(new_iiwu_thread): New structure and
2002-12-14 Peter Hanappe <firstname.lastname@example.org>
* src/iiwu_chan.c (iiwu_channel_cc): Handling NRPN messages (NRPN
* src/iiwu_voice.c (iiwu_voice_update_param): Does more extensive
range checking because the NPRN system may produce out-of-range
values (NRPN system).
(iiwu_voice_set_param): New function to change
generator values (NRPN system).
* src/iiwusynth_priv.h (iiwu_clip): New macro
* src/iiwu_synth.c (iiwu_synth_set_gen): New function to change
generator values (NRPN system).
* src/iiwu_gen.c (iiwu_gen_map_nrpn): New function to map the NRPN
data input to the parameter range (NRPN system).
* src/iiwu_midi.c (iiwu_midi_file_read_event): Fixed metadata
buffer bug (alloc size 1 too small).
2002-12-10 Peter Hanappe <email@example.com>
* src/iiwu_dsound.c (iiwu_win32_destroy_window): Filled in the
* src/iiwusynth.h: Changes in the definition for iiwu_synth_sfload
and iiwu_synth_sfunload, New functions: iiwu_synth_sfreload,
iiwu_synth_get_sfont_by_id, and iiwu_list_insert_at. New 'id'
field in iiwu_font_t.
2002-12-08 Markus Nentwig <firstname.lastname@example.org>
* src/Makefile.am: added iiwu_hash.c and iiwu_strtok.c to libiiwusynth_la_SOURCES
* src/iiwu_settings.c (iiwu_settings_init): Removed
multi_channel from the settings (replaced with audio_channels > 1)
* src/iiwu_settings.c (iiwu_settings_init): added audio_groups
setting. This is the number of individual channels generated from
the synth, and always equal to audio_channels, as long as the
LADSPA Fx unit is disabled. Otherwise it can be used (for
example) to separate even and odd MIDI channels, apply different Fx
and mix together to one stereo output.
src/iiwu_ladspa.c: Extended Fx unit to multigroup input, fx sends
and multiple audio output channels
2002-12-04 Peter Hanappe <email@example.com>
* src/iiwu_midi.c (iiwu_midi_file_read_event): the metadata buffer
is now dynamically allocated. What! Dynamic memory management
already existed in the sixties!
* src/iiwu_cmd.c (iiwu_handle_reset): New shell command. Sends
* src/iiwu_cmd.c (iiwu_expand_path): New function to handle
filenames starting with '~'.
* src/iiwu_cmd.c: Added commands for working with tunings. Added
* src/iiwu_chan.h (struct _iiwu_channel_t): added tuning
* src/iiwusynth.h: new tuning functions
* src/iiwu_synth.c (iiwu_synth_reset_tuning): new tuning functions
* src/iiwu_voice.c: Added tuning
2002-12-03 Peter Hanappe <firstname.lastname@example.org>
* doc/iiwusynth.1: new man page
* src/iiwu_midi.c (iiwu_player_load): the player now handles a
* src/iiwusynth.h: 'iiwu_player_add' replaces 'iiwu_player_load'
* src/iiwusynth.c (main): iiwusynth can now play midifiles.
2002-12-02 Peter Hanappe <email@example.com>
* src/iiwu_sys.c (new_iiwu_timer): New argument 'auto_destroy' to
specify whether the timer should delete it's structure when the
timer is finished.
* src/iiwu_synth.c (iiwu_synth_sfunload): If the soundfont can not
be unloaded immediately, a timer thread is spinned of to unload it
later. On MacOS 9, the unload is tried at a subsequent 'load' or
* src/iiwusynth.h (struct _iiwu_sample_t): Added 'refcount' field
to test when a soundfont can be unloaded.
* src/iiwu_synth.c (iiwu_synth_nwrite_float): New function
allowing multi-channel audio output.
(iiwu_synth_init): Fixed 'amount' for pan. Now set to 500.
* src/iiwu_cmd.c (iiwu_synth_cmdshell): Added little prompt.
* src/iiwusynth.c (print_welcome): iiwusynth prints out a welcome
message as an well-behaved, interactive application should.
* src/iiwu_synth.c (iiwu_synth_all_sounds_off): New function to
implement the 'All Sound Off' MIDI messages (CC 120).
(iiwu_synth_system_reset): This function now also resets the
default controller values on the MIDI channels, and clears the
reverb and chorus delay lines.
(iiwu_synth_count_midi_channels): New function to retreive the
number of available midi channels.
(iiwu_synth_count_audio_channels): New function to retreive the
number of available midi channels.
(iiwu_synth_count_effects_channels): New function to retreive the
number of available effects channels.
(iiwu_synth_get_cpu_load): New function to retreive an estimation
of the CPU load.
* src/iiwusynth.h: Added fields to handle multi-channel audio and
a variable number of midi-channels. The 'flags' has been
expanded/replaced with several variables.
* src/iiwu_chan.c (iiwu_channel_cc): Implemented the 'All Control
Off' MIDI message (CC 121).
* src/iiwu_chorus.c (iiwu_chorus_update): iiwu_chorus_update
(called after the iiwu_chorus_set_xxx function) no longer returns
an error of out-of-range values. It clips the value the the
2002-11-22 Markus Nentwig <firstname.lastname@example.org>
* src/iiwu_voice.c (iiwu_voice_write): Fixed compilation problem
without --enable-SSE (Pentium II and Mac)
2002-11-17 Markus Nentwig <email@example.com>
* src/iiwu_voice.c (iiwu_voice_write): Fixed nonlooped samples-bug.
* TODO (TODO): Updated
* src/iiwu_cmd.c (iiwu_handle_reverbsetlevel): Changed command
line command 'rev_setwet' to 'rev_setlevel'. Replaced the word 'wet' by
'level' in most places.
Added a command line option --dump, which provides
'machine-readable' output from stdout to hook up a user interface.
* src/iiwusynth.h: Moved the default values for gain, chorus and
reverb here. Might be useful as an example...
* src/iiwu_voice.c (iiwu_voice_calculate_runtime_synthesis_parameters):
Added 'scale tuning' modulator, centered around C3.
* src/iiwusynth.h: Added API functions to read the reverb state
Moved iiwu_synth_system_reset to the API
2002-11-08 Markus Nentwig <firstname.lastname@example.org>
* src/iiwu_voice.c (iiwu_voice_write): Fixed Volume envelope delay bug
* src/iiwu_voice.c (FILTER_TRANSITION_SAMPLES): Doubled filter
* src/iiwu_mod.c (iiwu_mod_get_value): Changed convex unipolar
* src/iiwu_voice.c (iiwu_voice_off): Cleaned up a bit, uses now
calls to iiwu_voice_off, when a voice is finished.
* src/iiwu_midi.c (iiwu_midi_parser_parse): Reimplemented
New parser should be able to cope with realtime, system common
2002-10-31 Markus Nentwig <email@example.com>
* src/iiwu_alsa.c (iiwu_alsa_midi_run): Increased MIDI timeout
from 1 to 100 ms
* src/iiwu_dsp_core.c: Merged identical filter coefficients b0 and
b2 into b02
Implemented smooth filter transitions
* src/iiwu_sys.c (iiwu_check_fpe): Added verbose FPE reporting and
systematic FPE checks.
* src/iiwu_rev.c: Added a constant DC offset to avoid slowdown
caused by denormal numbers
* src/iiwu_synth.c (delete_iiwu_synth): Fixed segv during shutdown
* src/iiwu_dsp_core.c: Fixed buffer bug (aligned-unaligned)
* src/iiwu_synth.c (iiwu_synth_damp_voices): Commented out unused code
* src/iiwu_dsp_core.c: Optimized, added SSE code, which is
this time actually faster than the default code. Well. Part of it.
* src/iiwu_voice.c: Minor clean-up
* configure.in: Added switch --enable-longlong
* configure.in: Added switch --enable-SSE
* src/iiwu_phase.h: Added 64 bit operations, documented
* src/iiwu_sse.h: Check to avoid #including the file more than once
2002-10-29 Markus Nentwig <firstname.lastname@example.org>
* src/iiwu_voice.c: Added experimental SSE support for Pentium
III. Comment out #define SSE from iiwu_voice.c to get back to the standard version.
2002-10-26 Markus Nentwig <email@example.com>
* src/iiwu_seq.c: Fixed a couple of warnings
* src/iiwu_voice.c (new_iiwu_voice): Removed iiwu_voice_init.
* src/iiwu_dsp_core.c: New 7th order interpolation.
2002-10-24 Markus Nentwig <firstname.lastname@example.org>
Added checking for invalid sample.
* src/iiwu_voice.c (iiwu_voice_write):
Moved the DSP core functions into iiwu_dsp_core.c.
Optimized, cleaned up, documented.
Amplitude scaling short => floating point is now done as the last
operation in the DSP loop (voice->amp does not include the scaling
* src/iiwu_synth.c (iiwu_synth_one_block): Saved a couple of
multiplications per sample by moving the master gain into iiwu_voice_write
* src/iiwu_synth.c (iiwu_synth_free_voice_by_kill): Modified the algorithm
* src/iiwu_synth.c (iiwu_synth_alloc_voice): Noteon algorithm will
now turn off retriggered running voices ('sustain pedal problem')
2002-10-18 Markus Nentwig <email@example.com>
* src/iiwu_alsa.c (new_iiwu_alsa_midi_driver): Disabled
high-priority scheduling for the MIDI thread to get rid of audio dropouts.
* src/iiwu_synth.c (iiwu_synth_free_voice_by_kill): Modified voice
killing algorithm, so that recently started voices are not killed
* src/iiwu_voice.c (iiwu_voice_run_dsp): Changed some variable
names. Extensive loop point checking, when loop points are modulated.
* src/iiwusynth.h: Added functions to read the state of the
* src/iiwu_chorus.c: Rewrote chorus setup logic (if a parameter is
out-of-range, all other parameter changes are discarded).
* src/iiwu_voice.c: Added caching for loop peak detection: The
amplitude of the loop is only detected once for each
sample. Exception only, if the resulting loop differs from the
original loop settings of the sample (in this case, the peak
detection is still run for each noteon event).
* src/iiwusynth.h (struct _iiwu_sample_t): Added 'amplitude_that_reaches_noise_floor_is_valid'
2002-07-21 Peter Hanappe <firstname.lastname@example.org>
* src/Makefile.am (libiiwusynth_la_SOURCES): Followed Bob Ham's
suggestion for the Makefile.am to fix the problems with automake
1999-11-30 Tim Goetze <email@example.com>
* src/iiwu_synth.c (iiwu_synth_alloc_voice): New algorithm for
voice allocation, when all voice processes are in use
1999-11-30 Markus Nentwig <firstname.lastname@example.org>
* src/iiwu_synth.c (iiwu_synth_alloc_voice): Applied above patch,
2002-07-08 Markus Nentwig <email@example.com>
* src/iiwu_synth.c (iiwu_synth_noteoff): Changed noteoff strategy:
Noteoff now turns off all voice processes with the same channel /
key, regardless of the voice ID (avoids stuck notes).
2002-07-13 Peter Hanappe <firstname.lastname@example.org>
* src/Makefile.am (EXTRA_libiiwusynth_la_SOURCES): Applied Takashi
Iwai's patch. The configure stuff in iiwusynth-0.2 cannot be
rebuilt with the latest automake 1.6. You cannot use substitution
for *_SOURCES in Makefile.am. This fixes this problem.
1999-11-30 Markus Nentwig <email@example.com>
* src/iiwusynth.h: Added documentation, removed GEN_CHANGED (it
* src/iiwu_mod.c (iiwu_dump_modulator): Cleaned up
* src/iiwu_cmd.c (iiwu_handle_help): Restructured command line
2002-06-14 Markus Nentwig <firstname.lastname@example.org>
* src/iiwu_chorus.c (iiwu_chorus_processmix): Turning off chorus now, when parameters are wrong (avoid FPE)
* src/iiwu_voice.c (iiwu_voice_write): Optimized turnoff condition for voice
2002-06-11 Markus Nentwig <email@example.com>
* src/iiwu_voice.c (iiwu_voice_add_mod): Fixed bug that prevented non-default modulators from being added.
(iiwu_voice_config): Added peak detection for the sample loop, and a condition turning off the voice,
if loop peak volume and amplitude envelope combined fall below the noise floor.
2002-06-06 Peter Hanappe <firstname.lastname@example.org>
* acinclude.m4: Fixed problems with enable/disable jack and
2002-06-06 Tim Goetze <email@example.com>
* src/iiwu_synth.c (iiwu_synth_all_notes_off): Added handling of
all-notes-off midi message
2002-06-03 Markus Nentwig <firstname.lastname@example.org>
* src/iiwu_chorus.c: Fixed bug in initial phase calculation
2002-06-02 Peter Hanappe <email@example.com>
* src/iiwu_jack.c: updated for new JACK types.
2002-06-02 Bob Ham <firstname.lastname@example.org>
* acinclude.m4: Changed acinclude.m4 for configure to ignore jack.
2002-06-02 Markus Nentwig <email@example.com>
* autogen.sh: Added libtoolize -f to prevent error message 'libtool: ltconfig version does not match ltmain.sh version ...'
* src/iiwusynth.h: Changed iiwu_voice_add_mod_t to iiwu_voice_add_mod
* src/iiwu_synth.c: Added NULL termination to list returned by iiwu_synth_get_voicelist
* src/iiwusynth.h: Added iiwu_synth_set_chorus (API function)
* src/iiwu_synth.c: Added iiwu_synth_set_(reverb|chorus)_on (API functions)
* src/iiwu_cmd.c: Added control commands for chorus (see help)
2002-05-26 Tim Goetze <firstname.lastname@example.org>
* src/iiwu_voice.c (iiwu_voice_noteoff): Fixed conversion between
volenv-values from attack segment to later envelope segments
2002-05-22 Markus Nentwig <email@example.com>
* src/iiwu_voice.c (iiwu_voice_query_ID): Added, API function
(iiwu_voice_query_playing): Added, API function
(iiwu_voice_write): Fixed problem with filter caused 05-18
* src/iiwusynth.h: Moved iiwu_voice_update_param into the API
2002-05-19 Markus Nentwig <firstname.lastname@example.org>
* src/iiwusynth.h (iiwu_synth_get_voicelist): Added.
* src/iiwu_voice.c (iiwu_voice_noteoff): Added a conversion for linear to cB amplitude, when a note is turned off
during the attack phase of the volume envelope
* src/iiwu_gen.h: Moved the generator definition to API. Changed the fields to 'double'.
* src/iiwu_mod.c: Moved the modulator definitions to API. Changed the data type of amount to 'double'.
* src/iiwu_voice.c (iiwu_voice_write): The condition, that quits a voice, when the amplitude
falls below a threshold now uses only the volume envelope instead of the voice amplitude.
Previously, turning a volume pedal briefly to 0 would quit all voices playing.
* src/iiwu_rev.c (iiwu_revmodel_processreplace): Removed 'dry' path from reverb unit
Motivation: This saves a couple of multiplications, the dry signal goes through the ordinary output
* src/iiwusynth.h (iiwu_synth_kill_by_exclusive_class): added to API
* src/iiwu_synth.c (iiwu_synth_kill_by_exclusive_class): Extended the exclusive class function to
work with stereo samples
(iiwu_synth_set_reverb): Renamed iwu_synth_set_reverb to iiwu_synth_set_reverb_preset
iiwu_synth_set_reverb is now an API function, that allows to set all reverb parameters.
2002-05-18 Markus Nentwig <email@example.com>
* src/iiwu_chorus.c: Implemented variable delay line with bandlimited interpolation. Documentation, error handling.
Removed unneeded and broken features
* src/iiwusynth_priv.h: Moved typedef struct iiwu_mod_t iiwu_mod_t into iiwusynth.h
* src/iiwusynth.h: Moved iiwu_voice_add_mod from iiwu_voice.h into iiwusynth.h (now API function).
* src/iiwu_voice.c (iiwu_voice_update_param): Inserted chorus send into DSP loop
(iiwu_voice_write): Added flag 'voice->update_filter'. Now Q can be modulated.
2002-05-12 Markus Nentwig <firstname.lastname@example.org>
* src/iiwu_synth.c (iiwu_synth_pitch_wheel_sens): added
* src/iiwu_chan.c (iiwu_channel_pitch_wheel_sens): added
* src/iiwu_cmd.c (iiwu_handle_reverbsetwidth): changed 'wet' to
2002-05-11 Markus Nentwig <email@example.com>
* src/iiwu_conv.c (iiwu_tc2sec):
Added more conversion functions with range check for different ranges:
* src/iiwu_voice.c (iiwu_voice_add_mod):
implemented modulator src 0 (constant mod offset)
* src/iiwu_voice.c (iiwu_voice_update_param):
sample-and envelope related voice parameters
are now handled together with other voice parameters.
2002-05-10 Peter Hanappe <firstname.lastname@example.org>
* src/iiwu_synth.c (iiwu_synth_start_voice): added
iiwu_synth_start_voice() to handle exclusive classes.
2002-05-09 Peter Hanappe <email@example.com>
* src/iiwu_conv.h: removed velocity to cB conversion. No longer
* src/iiwu_synth.c (iiwu_synth_write_float): removed limiter
* src/iiwu_synth.h (IIWU_NUM_CHANNELS): set the number of channels
* src/iiwu_synth.c (iiwu_synth_get_internal_bufsize): added
* src/iiwu_ladspa.h: lower-cased ladspa files
* src/iiwusynth.h: prefixed log levels with IIWU_... Updated all
* src/iiwu_cmd.c (iiwu_handle_reverb): renamed 'rev_enable' to
'reverb' in correspondance with the long command line arguments
* src/iiwusynth.c (main): checking if files on command line are valid
* src/iiwuplay.c (main): checking if files on command line are valid
* src/iiwusynth.h: New log level for verbose messages: IIWU_INFO
2002-04-30 Markus Nentwig <firstname.lastname@example.org>
* src: Added iiwu_LADSPA.c, iiwu_LADSPA.h (support for LADSPA effect plugins).
* src/iiwusynth.c (main): Changed default gain to 0.2.
* src/iiwu_voice.c: Restructured the voice initialization as follows:
(iiwu_voice_init): sample position, IIR filter history, envelopes etc. are reset.
(iiwu_voice_optimize): The generators (nominal value) have been set by the sound font.
Now each modulator is calculated once to obtain the 'final' initial value for each
generator, which consists of nominal value and modulator-contributed part.
(iiwu_voice_update_param): Calculates all voice parameters, which depend on one
particular generator. This is called once for each voice parameter during
voice_optimize and further each time, when a modulator changes a generator.
(iiwu_voice_update_param): Added a voice parameter filter_gain to avoid
recalculating the filter gain each time the center frequency changes (it depends
only on Q)
(iiwu_voice_write): Voice is now turned off, when the amplitude falls below -100 dB,
even during the sustain phase (happens, when holding a piano key for a very long time)
* src/iiwu_voice.c (iiwu_voice_noteoff): Moved voice->chan = NO_CHANNEL into iiwu_voice_off.
Previously a released note was not modulated anymore, for example pitch bend stopped working as soon
as the key was released.
* src/iiwu_voice.h: Changed _ON macro to figure out the state of a key from the position in
the envelope, instead of using a cleared channel number as indicator.
* src/iiwu_synth.c: Implemented all default modulators
Added LADSPA support.
Added digital clipping.
Moved master gain factor ahead of LADSPA Fx.
* src/iiwu_mod.c: 'Hardcoded' GM default modulator vel => filter.
Replaced 128 with 127 in (127-x)
* src/iiwu_midi.c: Fixed sysex for realtime MIDI. Fixed pitch bend bug.
* src/iiwu_gen.c (iiwu_gen_set_default_values): Using float instead of int for default values.
Added references to specifications (doc). Changed 'init array' function name to 'set_default' .
* src/iiwu_defsfont.c (iiwu_preset_zone_import_sfont): Import of modulators
(iiwu_inst_zone_import_sfont): Import of modulators
(iiwu_defpreset_noteon): Added modulators, fixed generator problem (local zone overwrites global zone, previously it added)
* src/iiwu_conv.c: Using now oncave / convex equation from SF specs.
Removed ct2hz functions and tables.
(iiwu_ct2hz): Limit checking
(iiwu_cb2amp): Removed 'magic number'
(iiwu_tc2sec): Avoided == for iiwu_real_t
* src/iiwu_cmd.c: Increased number of tokens. Using WORKLINELENGTH constant.
Changed max. gain to 5.
Added LADSPA commands.
Renamed misleading rev_bypass command to rev_enable
* src/iiwu_chan.c: Centered pitch wheel. Added 'expression' modulator (CC 11).
* configure.in: Added LADSPA support
2002-04-03 Peter Hanappe <email@example.com>
* src/iiwu_voice.c (iiwu_voice_run_dsp): Integrated Markus
Nentwig's new filter design
2002-03-12 Peter Hanappe <firstname.lastname@example.org>
* src/iiwusynth.h: the preset iteration in a soundfont now takes a
pointer to a preset structure
* src/iiwu_sys.c (iiwu_profile_data): added support for profiling
* src/iiwu_voice.c (iiwu_voice_write): turns off voice if
amplitude < -100 dB in release phase. Set filter gain back to old
value (0.25f * ...)
* src/iiwuplay.c (main): added gain, interactive, and reverb options
* src/iiwusynth.c (main): added gain and reverb options
* src/iiwu_synth.c (iiwu_synth_write_s16): added brickwall limiter for s16 samples
2002-01-29 Stephane Letz <email@example.com>
* src/iiwu_midishare.c : Compilation on MacOSX, use a task for typeNote management
* src/iiwu_sys.c : Compilation on MacOSX
* src/iiwu_sys.h : Compilation on MacOSX
* src/iiwu_sfont.c : Use the flag MACINTOSH instead of MACOS
* config_macos.h : Cleanup
* config_macosx.h : New file, compilation on MacOSX
2002-01-21 Stephane Letz <firstname.lastname@example.org>
* src/iiwu_midi.c (delete_iiwu_midi_handler): Desallocation of heap allocated strings
* src/iiwusynth_priv.h : Definition of strdup if not available (Macintosh)
2002-01-16 Peter Hanappe <email@example.com>
* src/iiwu_alsa.c (new_iiwu_alsa_seq_driver): Applied and adjusted
Bob Ham's patch: support for configurable ALSA sequencer client
* src/iiwu_chan.c (iiwu_channel_cc): Applied Bob Ham's patch:
added bank select midi message.
2001-12-31 Peter Hanappe <firstname.lastname@example.org>
* src/iiwu_synth.c (iiwu_synth_damp_voices): Sustain messages are
now handled. Updated iiwu_channel and iiwu_voice.
(delete_iiwu_synth): SoundFonts are deleted.
2001-12-21 Stephane Letz <email@example.com>
* src/iiwu_midishare.c (new_iiwu_midishare_midi_driver,
Updated to be compiled either in driver or application mode
with the flag MIDISHARE_DRIVER.
2001-12-20 Stephane Letz <firstname.lastname@example.org>
* src/iiwu_portaudio.c (iiwu_portaudio_run , new_iiwu_portaudio_driver):
Adaptation for new audio drivers
* src/iiwu_synth.c (audio driver definition):
Adaptation for PortAudio driver
* src/iiwu_sys.c (header): Adaptation for compilation on MacOS9
* src/iiwu_sys.h (header): Adaptation for compilation on MacOS9
2001-12-16 Peter Hanappe <email@example.com>
* src/iiwuplay.c (main): The .iiwusynth file is loaded *before*
the soundfonts on the command lines are loaded
* src/iiwusynth.c (main): idem.
2001-12-16 Peter Hanappe <firstname.lastname@example.org>
* src/iiwu_midi.c (iiwu_player_callback): Fixed error in midi
timing after a tempo change
* src/iiwu_jack.c (new_iiwu_jack_audio_driver): Added first
version of JACK driver
2001-12-14 Peter Hanappe <email@example.com>
* src/iiwu_synth.c (iiwu_synth_noteoff): noteon/notoff events can
print a clear message, useful for debugging.
* src/iiwu_sys.c (struct _iiwu_timer_t ): timer moved from
iiwu_midi.c to iiwu_sys.c
* src/iiwusynth.h: New organization of settings; using bit flags;
added verbose option
* src/iiwusynth.c (main): Added the verbose option
* src/iiwuplay.c (main): Added the verbose option
2001-10-05 Stephane Letz <firstname.lastname@example.org>
* src/iiwu_portaudio.c (new_iiwu_portaudio_driver): imported new
driver for the PortAudio library.
2001-10-04 Stephane Letz <email@example.com>
* src/iiwu_synth.c (new_iiwu_synth): Fixed bug in synth initialisation
2001-10-02 Peter Hanappe <firstname.lastname@example.org>
* src/iiwu_cmd.c (iiwu_get_userconf): returns default user
(iiwu_get_sysconf): returns default system configuration
(iiwu_synth_cmdline): Fixed bug with argument offset. Empty lines
are skipped correctly.
* src/iiwusynth.c (main): loads the user or system config
* src/iiwuplay.c (main): loads the user or system config
* src/iiwu_synth.c (iiwu_sp_write): Using new envelope model for
2001-09-29 Peter Hanappe <email@example.com>
* src/iiwu_synth.c (iiwu_sp_write): redesigned the envelopes.
2001-09-20 Peter Hanappe <firstname.lastname@example.org>
* src/iiwu_synth.c (iiwu_sp_write): redesigned the dsp loop. it's
faster and it sounds better (!)
2001-09-19 Peter Hanappe <email@example.com>
* src/iiwu_sfont.c (iiwu_sample_import_sfont): better checking for
minimum sample size, loop start and loop end offsets.
2001-09-17 Peter Hanappe <firstname.lastname@example.org>
* src/iiwu_synth.c (iiwu_sp_write): improved calculation of filter
(new_iiwu_synth): using settings structure
2001-09-09 Peter Hanappe <email@example.com>
* src/iiwu_synth.h (iiwu_phase_decr): fixed bug
* src/iiwu_synth.c (iiwu_synth_noteoff): noteoff now turns off the
oldest note only (instead of all notes with the given channel and
2001-07-10 Peter Hanappe <firstname.lastname@example.org>
* src/iiwu_midi.h: removed midi driver join function. updated all
structures, implementations and callers.
2001-07-04 Peter Hanappe <email@example.com>
* src/iiwuplay.c (print_help): corrected errors in the help and
2001-06-29 Peter Hanappe <firstname.lastname@example.org>
* src/iiwu_synth.c (iiwu_synth_one_block): new function. fills the
buffer with fresh samples.
(iiwu_synth_write_lr): now calls iiwu_synth_one_block. the
synthesizer uses fixed synthesis buffer size, independent of the
requested buffer length passed to iiwu_synth_write_lr.
(iiwu_revmodel_processreplace): new uses fixed IIWU_BUFSIZE value
for buffer length.
(iiwu_revmodel_processmix): uses fixed IIWU_BUFSIZE value
2001-06-22 Peter Hanappe <email@example.com>
* src/iiwusynth.c (iiwu_handle_fonts): new shell command to list
the loaded fonts.
(iiwu_handle_mstat): new shell command to list the statistics of
the midi driver.
2001-06-19 Peter Hanappe <firstname.lastname@example.org>
* src/iiwusynth.c (main): Several command line options are
available to select the midi and audio driver and device. Using
the getopt function on posix machines.
2001-06-16 Peter Hanappe <email@example.com>
* src/iiwu_synth.h: new iiwu_revmodel_presets_t structure to store
reverb presets (concert hall, room, ...)
* src/iiwu_synth.c (iiwu_synth_write_lr): now using 1 reverb for
all synthesis processes. the synthesis processes now receive a
left and right buffer, a reverb buffer, a chorus buffer, and a
monobuffer for their temporarry storage. reverb now always on.
(new_iiwu_sp): no longer allocating a reverb module nor a
monobuf. only one reverb model and monobuffer allocated by the
synth object (read: much less memory usage).
* src/iiwu_midi.c (iiwu_player_callback): fixed timing
errors. midi should play correctly now.
2001-06-09 Peter Hanappe <firstname.lastname@example.org>
* src/iiwu_synth.c (iiwu_sp_write_lr): now using a 64-bits
fixed-point number to calculate the phase of the
wavetable. because of rounding erros, the float value I used
before gave terrible tuning problems. I updated all the
* src/iiwusynth_priv.h: included the iiwu_phase_t data type. This
type represents a 64-bits fixed-point number. It's used to hold
the phase in the wavetable.
2001-06-08 Peter Hanappe <email@example.com>
* src/iiwu_midi.c (new_iiwu_midi_handler): Better support for
runtime selection of the MIDI driver (using the
* src/iiwu_auport.c (new_iiwu_auport): Better support for runtime
selection of the audio driver (using the iiwu_adriver_definition_t
2001-06-07 Peter Hanappe <firstname.lastname@example.org>
* src/iiwu_synth.c (iiwu_sp_write_lr): rewrote the dsp function to
accept a seperate left and right channel buffer.
(iiwu_sp_write_lr): using cubic hermite interpolation by default.
(iiwu_synth_write_lr): added a dsp function to accept a seperate
left and right channel buffer.
2001-05-26 Peter Hanappe <email@example.com>
* src/iiwu_midi.c (iiwu_midi_parser_parse): Fixed a bug in the
midi parser (running status should not be split in a status and
channel part for system messages).
(iiwu_midi_send_event): pitch bend events are now handled
2001-05-25 Peter Hanappe <firstname.lastname@example.org>
* src/iiwu_midi.c (iiwu_midi_file_getc): Fixed bug when pushed
back byte equals zero (mf->c >= 0)
* src/iiwu_midi.c (iiwu_midi_file_getc): Fixed bug when pushed
back byte equals zero (mf->c >= 0)
2001-05-24 Peter Hanappe <email@example.com>
* src/iiwusynth.c: added the stupidly simple interpreter
* src/iiwu_synth.c: removed all param strcutures.
* src/iiwu_synth.c (iiwu_channel_get_banknum): new function
2001-05-23 Peter Hanappe <firstname.lastname@example.org>
* src/iiwu_synth.c (iiwu_sp_write): Fixed devide by zero in filter
* src/smurf.c (gerr): applied Josh's patch: using va_list now (as
2001-05-22 Peter Hanappe <email@example.com>
* src/iiwu_midi.c: the midi handler is now devided in a dummy
iiwu_midi_handler_t and a "low level" driver. This allows for
multiple midi drivers to be compiled in.
* src/iiwusynth.h: renamed iiwu_midi_driver_t to iiwu_midi_handler_t
* src/iiwu_auport.c (new_iiwu_auport): new "driver" argument to
select between alsa, oss, midishare, directx, ...
* configure.in: preparing for the first pre-release, version 0.0.1