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, see: http://sourceforge.net/apps/trac/fluidsynth/wiki/ChangeLog1_1_2 http://sourceforge.net/apps/trac/fluidsynth/wiki/ChangeLog1_1_1 etc. For developer related "What's new"-information, doc/fluidsynth-v11-devdoc contains valuable information. === OLD === 2009-05-01 Pedro Lopez-Cabanillas * configure.ac: fix for win32 build. 2009-05-01 Pedro Lopez-Cabanillas * 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 * 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 config_win32.h.in. 2009-04-26 Josh Green * 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 being ignored. * 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 * 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 * src/config_win32.h: fix compilation under MSVC 2008 and older 2009-03-15 Josh Green * ltconfig: Removed obsolete ltconfig script by suggestion of Sven Hoexter. * 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 * 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 * 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 * Applied patch from KO Myung-Hun for OS/2 support including Dart audio driver. 2009-01-29 Josh Green * 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 * 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 * 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 * 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 * src/fluid_dsound.c: Fix for ticket #16 - dsound device can't be selected. 2008-09-07 Josh Green * 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 midi.winmidi.device setting. * 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 setting. (fluid_synth_program_change): added fix to properly search for a percussion instrument * 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 midi.winmidi.device option. * 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 * 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 * 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 longer needed. 2007-11-10 Josh Green * 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 * 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 reverb level. * 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 * 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 see http://fluidsynth.resonance.org/trac/ticket/2. * : Applied reverb damp scaling patch from David Hilvert see http://fluidsynth.resonance.org/trac/ticket/3. * 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 * 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 * 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 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 tokenizing instance. * 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 handling, * 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 other improvements. * src/fluid_synth.c (delete_fluid_synth): Turning off all voices so that SoundFont data will be freed correctly (thanks to patch from Paul Millar). * 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 * 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 type settings. * src/fluid_synth.c: Memory leak fixes: Freeing FX buffers and right/left_buf. 2006-11-21 Josh Green * 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 * 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 * 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 * Fixed build error that occured when neither LASH or LADCCA are present. * Updated README-OSX from Ebrahim Mayat. 2006-02-18 Josh Green * 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 * src/fluid_ramsfont.c (fluid_ramsfont_remove_izone): Applied crash bug fix from Antoine Schmitt. 2005-07-05 Josh Green * 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 * Applied LASH patch that is included with ladcca-0.4.0. 2005-06-11 * 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 currently. * src/Makefile.am: Now using LT_VERSION_INFO to substitute the libtool version. * 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. 2005-06-11 * 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 initial attenuation. (fluid_voice_noteoff): Re-coded volenv_val attack conversion and verified. 2005-06-10 * 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. 2005-06-07 * 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 FLUID_CB_POWER_FACTOR. 2004-11-11 * README-OSX: Update from Ebrahim Mayat. 2004-08-18 * 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). 2004-08-06 * 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! 2004-07-29 * 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 2004-05-14 * doc/fluidsynth.1 (option): Fixed typo noted by Gerald Pye. 2004-05-14 Peter Hanappe * 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 * src/fluid_synth.c (fluid_synth_remove_sfont): Added new function 2004-05-05 Peter Hanappe * 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 anti-aliasing filter. 2004-05-04 Peter Hanappe * 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 * 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 * 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 * src/fluid_jack.c (new_fluid_jack_audio_driver2): Testing the number of ports before allocating them. (fluid_jack_audio_driver_settings): Registering 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 below). * src/fluid_strtok.c (fluid_strtok_char_index): Removed printf's from fluid_strtok.c 2004-03-26 Peter Hanappe * bindings/README: Imported the fluidsynth_jni and fluidmax projects. 2004-03-25 Peter Hanappe * 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 * 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 * 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 * 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 * 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 * 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 * 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 * 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 mutex. 2003-11-14 Josh Green * src/fluidsynth.c: Removed CCA_Use_Jack and CCA_Use_Alsa flags since LADCCA no longer uses them. 2003-08-31 Josh Green * 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 source files. * Makefile.am: Re-arranged SUBDIRS so build output looks nicer. * src/Makefile.am: ALSA and OSS are now conditionally compiled using automake conditionals. 2003-08-29 Josh Green * 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 FLUID_REALLOC macro. * 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 * 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 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 time. * Makefile.am: Added fluidsynth.spec(.in) to EXTRA_DIST. 2003-08-19 Josh Green * 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 update routine. * 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 * 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 * 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 be lower?). * 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 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 * configure.ac: Fixed detection of CoreAudio by looking for CoreAudio/AudioHardware.h. * 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 script though). * fluidsynth_priv.h: Added "#define WITHOUT_SERVER 1" to Darwin build. 2003-06-12 Josh Green * 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 Alexandre Prokoudine. 2003-06-09 Josh Green * 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 * 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 * 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 simply disabled). 1999-11-30 Antoine Schmitt * src/fluid_defsfont.c: inst_zone lokey is now properly inialized to 0 (it was not, leading to random lost noteons depending on memory initialization) 2003-04-03 Peter Hanappe * src/fluid_rev.c: reverb parameters are clipped to their valid range. * 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 pitchbend value 2003-02-27 Josh Green Updated automake files (automake 1.6). * configure.ac: New version autoconf variables which get substituted into include/iiwusynth/version.h.in. * include/iiwusynth/version.h.in: Version defines that are filled in by autoconf. * 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 * src/iiwu_ladspa.c: Added a very small signal at Nyquist frequency. This fixes denormal number problems in some plugins. * 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 autogen.sh. 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 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 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 client name. * 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 * 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 * 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 * 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 * 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 interface. (new_iiwu_cmd_handler): New structure and functions to improve command interface. * src/iiwu_sys.c (new_iiwu_server_socket): New structure and functions (new_iiwu_thread): New structure and functions 2002-12-14 Peter Hanappe * src/iiwu_chan.c (iiwu_channel_cc): Handling NRPN messages (NRPN system). * 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 * src/iiwu_dsound.c (iiwu_win32_destroy_window): Filled in the empty lines... * 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 * 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 * 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 system reset. * 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 'source' command. * 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 * doc/iiwusynth.1: new man page * src/iiwu_midi.c (iiwu_player_load): the player now handles a playlist. * src/iiwusynth.h: 'iiwu_player_add' replaces 'iiwu_player_load' * src/iiwusynth.c (main): iiwusynth can now play midifiles. 2002-12-02 Peter Hanappe * 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 'unload' request. * 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 [min-max] range. 2002-11-22 Markus Nentwig * src/iiwu_voice.c (iiwu_voice_write): Fixed compilation problem without --enable-SSE (Pentium II and Mac) 2002-11-17 Markus Nentwig * 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 * src/iiwu_voice.c (iiwu_voice_write): Fixed Volume envelope delay bug * src/iiwu_voice.c (FILTER_TRANSITION_SAMPLES): Doubled filter fading time * src/iiwu_mod.c (iiwu_mod_get_value): Changed convex unipolar negative definition * 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 and resynchronize. 2002-10-31 Markus Nentwig * 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 * 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 * 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 * src/iiwu_voice.c(iiwu_voice_determine_amplitude_that_reaches_noise_floor_for_sample): 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 factor anymore). * 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 * 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 chorus. * 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' and 'amplitude_that_reaches_noise_floor' * src/iiwu_voice.c(iiwu_voice_calculate_runtime_synthesis_parameters): Renamed 'iiwu_voice_optimize' 2002-07-21 Peter Hanappe * src/Makefile.am (libiiwusynth_la_SOURCES): Followed Bob Ham's suggestion for the Makefile.am to fix the problems with automake 1.6 1999-11-30 Tim Goetze * 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 * src/iiwu_synth.c (iiwu_synth_alloc_voice): Applied above patch, 2002-07-08 Markus Nentwig * 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 * 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 * src/iiwusynth.h: Added documentation, removed GEN_CHANGED (it was unused). * src/iiwu_mod.c (iiwu_dump_modulator): Cleaned up * src/iiwu_cmd.c (iiwu_handle_help): Restructured command line help system 2002-06-14 Markus Nentwig * 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 * 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 * acinclude.m4: Fixed problems with enable/disable jack and midishare 2002-06-06 Tim Goetze * src/iiwu_synth.c (iiwu_synth_all_notes_off): Added handling of all-notes-off midi message 2002-06-03 Markus Nentwig * src/iiwu_chorus.c: Fixed bug in initial phase calculation 2002-06-02 Peter Hanappe * src/iiwu_jack.c: updated for new JACK types. 2002-06-02 Bob Ham * acinclude.m4: Changed acinclude.m4 for configure to ignore jack. 2002-06-02 Markus Nentwig * 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 * src/iiwu_voice.c (iiwu_voice_noteoff): Fixed conversion between volenv-values from attack segment to later envelope segments 2002-05-22 Markus Nentwig * 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 * 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 anyway. * 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 * 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 * 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 * 'width' 2002-05-11 Markus Nentwig * src/iiwu_conv.c (iiwu_tc2sec): Added more conversion functions with range check for different ranges: (iiwu_tc2sec_attack): (iiwu_tc2sec_hold): (iiwu_tc2sec_release): * 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. Implemented generators: GEN_KEYTOVOLENVDECAY GEN_KEYTOVOLENVHOLD GEN_KEYTOMODENVDECAY GEN_KEYTOMODENVHOLD 2002-05-10 Peter Hanappe * src/iiwu_synth.c (iiwu_synth_start_voice): added iiwu_synth_start_voice() to handle exclusive classes. 2002-05-09 Peter Hanappe * src/iiwu_conv.h: removed velocity to cB conversion. No longer used. * src/iiwu_synth.c (iiwu_synth_write_float): removed limiter * src/iiwu_synth.h (IIWU_NUM_CHANNELS): set the number of channels to 64. * 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 references. * 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 * 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 * src/iiwu_voice.c (iiwu_voice_run_dsp): Integrated Markus Nentwig's new filter design 2002-03-12 Peter Hanappe * 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 * 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 * 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 * src/iiwu_alsa.c (new_iiwu_alsa_seq_driver): Applied and adjusted Bob Ham's patch: support for configurable ALSA sequencer client name. * src/iiwu_chan.c (iiwu_channel_cc): Applied Bob Ham's patch: added bank select midi message. 2001-12-31 Peter Hanappe * 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 * src/iiwu_midishare.c (new_iiwu_midishare_midi_driver, delete_iiwu_midishare_midi_driver): Updated to be compiled either in driver or application mode with the flag MIDISHARE_DRIVER. 2001-12-20 Stephane Letz * 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 * 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 * 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 * 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 * src/iiwu_portaudio.c (new_iiwu_portaudio_driver): imported new driver for the PortAudio library. 2001-10-04 Stephane Letz * src/iiwu_synth.c (new_iiwu_synth): Fixed bug in synth initialisation 2001-10-02 Peter Hanappe * src/iiwu_cmd.c (iiwu_get_userconf): returns default user configuration (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 modulation envelope 2001-09-29 Peter Hanappe * src/iiwu_synth.c (iiwu_sp_write): redesigned the envelopes. 2001-09-20 Peter Hanappe * src/iiwu_synth.c (iiwu_sp_write): redesigned the dsp loop. it's faster and it sounds better (!) 2001-09-19 Peter Hanappe * 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 * src/iiwu_synth.c (iiwu_sp_write): improved calculation of filter coefficients (new_iiwu_synth): using settings structure 2001-09-09 Peter Hanappe * 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 key) 2001-07-10 Peter Hanappe * src/iiwu_midi.h: removed midi driver join function. updated all structures, implementations and callers. 2001-07-04 Peter Hanappe * src/iiwuplay.c (print_help): corrected errors in the help and usage display. 2001-06-29 Peter Hanappe * 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 * 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 * 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 * 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 * 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 intepolation macros. * 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 * src/iiwu_midi.c (new_iiwu_midi_handler): Better support for runtime selection of the MIDI driver (using the iiwu_mdriver_definition_t structure) * src/iiwu_auport.c (new_iiwu_auport): Better support for runtime selection of the audio driver (using the iiwu_adriver_definition_t structure) 2001-06-07 Peter Hanappe * 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 * 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 * 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 * 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 * 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 it should). 2001-05-22 Peter Hanappe * 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