Commit graph

114 commits

Author SHA1 Message Date
Randy Heit
ebd115e7ca - Removed snd_3dspread, because it totally does not do what I want. I was using it to preserve
some of the stereoness of stereo sounds played in 3D. My testing was done only with stereo
  speakers, however, and I did not realize that it was moving the perceived physical location
  of the sound itself (because it sounded fine with my two speakers). So when 3D spread started
  working with mono sounds as well in FMOD 4.28, sound positioning was completely broken for
  everything when outputting to more than two speakers, because sounds were being spread
  across a 180 degree arc.

  Whoops!

  Stereo sounds are now completely mono when not played by you, the listener.

SVN r3357 (trunk)
2012-02-14 03:15:13 +00:00
Randy Heit
68fbd75897 - Fixed compilation with FMOD 4.38+. The removal of hardware voices and EAX was a fairly
major change, so I'm making no provisions for using older FMOD DLLs when compiled with the
  4.38 API. However, sound positioning is still broken like in 4.28, so you are recommended
  to continue building with 4.26. Also, the Freeverb-based DSP unit is no longer present in
  FMOD, so the underwater effect is currently unavailable when using 4.38 until I can figure
  out how to make it work with the SFX Reverb unit instead. (And on that topic, the Freeverb
  DSP was officially only for stereo outputs, so I really shouldn't have been using it in the
  first place.)
- Since I would like to eventually figure out the sound positioning issues with the newer
  FMODs, the following have been added:
  * snd_drawoutput now labels its outputs with the speakers they represent.
  * DCanvas::DrawTextA was added as an alias for DrawText, since the Windows headers #define
    DrawText to a Unicode/non-Unicode variant.
  * The loopsound console command was added to spawn an actor at the player's location and
    have it loop a sound infinitely.
  Hopefully I can figure it out. FMOD's 3D example works, so I assume the problem lies with
  my code, though I don't really know where to begin looking for the problem.

SVN r3350 (trunk)
2012-02-10 00:53:50 +00:00
Braden Obrzut
5c81cf20ba - Merged r1236, r1237, and r1240 of GZDoom.
SVN r3279 (trunk)
2011-08-14 23:53:20 +00:00
Braden Obrzut
f42358ad08 - Backported FMOD Ex 4.34 fixes from gzdoom-macosx. (With changes to continue to allow compilation with 4.22-4.28.)
SVN r3278 (trunk)
2011-08-13 21:11:09 +00:00
Christoph Oelckers
1993e969b7 - GCC warning fixes (GCC stupidity for the most part.)
SVN r3154 (trunk)
2011-02-19 08:59:43 +00:00
Randy Heit
869b9c9ccc - Fixed: FMODSoundRenderer::DrawWaveDebug() only allocated enough space for the wavearray based
on the window_size, so for large numbers of output channels, it would not allocate enough
  space for the spectrum data (which is definied by SPECTRUM_SIZE, not the window_size) and
  write junk on the stack when drawing the spectrums, causing a crash.

SVN r3087 (trunk)
2011-01-02 18:11:48 +00:00
Christoph Oelckers
5ff3b3f329 - added Gez's VOC loader.
SVN r2901 (trunk)
2010-10-05 22:31:22 +00:00
Randy Heit
2c397b1aac - Add a default search for Creative's CT4MGM.SF2 on Windows if fluid_patchset is unspecified.
- Use fluid_patchset to specify the SoundFont(s) for FluidSynth instead of doubling up with
  snd_midipatchset, which is already used by FMOD.

SVN r2555 (trunk)
2010-08-19 15:46:49 +00:00
Randy Heit
d53ef3f38a - Oops. These changes didn't belong in the repository.
SVN r2547 (trunk)
2010-08-15 20:05:35 +00:00
Randy Heit
1fa4bbf69c - Added FluidSynth support as snd_mididevice -5. Only tested with Linux. I will have
to try compiling it myself on Windows to see if it's really that slow or if
  Ubuntu just ships an unoptimized version, because performance is pretty pathetic
  when compared to the other options. (I understand that it's a complete SoundFont2
  renderer, so it is understandably slower than something like TiMidity++, but still.
  Does it really need to be around 10x slower? I played with the chorus, reverb, and
  interpolation settings, and none of them seemed to make much difference in
  performance.)



SVN r2545 (trunk)
2010-08-15 19:54:59 +00:00
Randy Heit
3416a7fc47 - Fix compilation with MinGW + w32api and clean up warnings.
SVN r2505 (trunk)
2010-08-11 03:56:31 +00:00
Randy Heit
c29639426f - Sounds that define a loop no longer play looped by default. They must be started with
CHAN_LOOP so that the higher level sound code knows they loop and can handle them accordingly.
- Added support for a LOOP_BIDI tag. Set it to "1", "On", "True", or "Yes" to use a
  bidirectional loop. This only works with sounds and not music, because music is streamed
  so does not support them.
- Extended custom loop support to work with samples as well as music.

SVN r2434 (trunk)
2010-07-15 23:07:41 +00:00
Randy Heit
99a2014ead - Added support for custom loop points for songs. This does not work with MP3 because of the
way MP3 obfuscates custom tags. Vorbis and FLAC are fine. (I could make it work with MP3,
  but you should be using Vorbis instead.) They are:
  * LOOP_START: Start time for the loop. If omitted, the song repeats from the beginning.
  * LOOP_END: End time for the loop. If omitted, the song loops at the end. (If you need to specify this, why aren't you using a shorter song.)
  You only need to specify one of these tags to set the custom loop. Naturally, you can set
  them both, as well. The format for each tag is the same:
  * If it contains a colon (:), it specifies by time. This may be of the form 00:00:00.00
    (HH:MM:SS.ss) to specify by play. Various parts may be left off. e.g. To start the loop
    at 20 seconds in, you can use ":20", 0:20", "00:00:20", ":20.0", etc. Values after the
    decimal are fractions of a second and accurate to one millisecond.
  * If you don't include a colon but just have a raw number, then it's the number of PCM
    samples at which to loop.
  * Any characters other than digits (0-9), colons (:), or a single decimal point for the
    seconds portion will result in the tag being ignored.

SVN r2424 (trunk)
2010-07-10 02:59:34 +00:00
Randy Heit
08b931006f - Reverted accidental change to fmodsound.cpp from revision 2273.
SVN r2282 (trunk)
2010-04-17 01:55:33 +00:00
Randy Heit
114412d5a2 - Search for the MUS header within the first 32 bytes of the song so that the music in
diescum.wad works.

SVN r2273 (trunk)
2010-04-06 03:12:32 +00:00
Randy Heit
7bd741a644 - Fixed compiling with FMOD Ex 4.30.00. I get no 3D sound effects, and it's crashing on me, too.
Am I doing something wrong?

SVN r2271 (trunk)
2010-04-04 17:48:26 +00:00
Christoph Oelckers
a6a303f438 - fix division by 0.
SVN r2243 (trunk)
2010-03-25 23:29:41 +00:00
Randy Heit
6c63ce51e4 - Fixed: The listener used a different coordinate system than sounds, which meant that sounds
were never made head-relative, so stereo effects would disappear if they were played by
  the listener.

SVN r2231 (trunk)
2010-03-18 21:54:24 +00:00
Randy Heit
740d52cf5b - Added support for Blood's SFX loop start markers.
SVN r2221 (trunk)
2010-03-18 04:26:22 +00:00
Christoph Oelckers
9aa1e20c77 - Added Blzut3's Solaris patch.
SVN r2070 (trunk)
2010-01-01 09:21:04 +00:00
Randy Heit
51eb6465a2 - Fixed two bugs in FMODSoundRenderer::HandleChannelDelay():
* Looping sounds that have been playing for a very long time, were evicted,
    and then were restarted need to have their positions clamped to lie
    within the bounds of the sounds. If we try to set a start position very
    far beyond the end, it will overflow inside FMOD and not work.
  * A start time of 0 is not actually valid and means the sound was never
    assigned a start time.
- The latter bug also reveals a problem with starting looped sounds evicted:
  They need to be assigned a start time so if they should have the opportunity
  to start later, they will be properly synchronized.


SVN r1987 (trunk)
2009-11-18 04:45:20 +00:00
Randy Heit
fb74d9b1ec - Added virtual status and audibility to the noise debug display.
SVN r1961 (trunk)
2009-11-04 02:07:39 +00:00
Randy Heit
50b0340a56 - Fixed: S_EvictAllChannels() must replace the channel's start time with its
position when evicting sounds, because restarting the sound system causes
  the DSP clock to restart at 0, so start times that were recorded before
  the reset are no longer applicable after the reset.
- Fixed: S_StopChannel() always set the channel's actor to NULL, eliminating
  origin information when resetting the sound system.


SVN r1949 (trunk)
2009-10-30 02:18:07 +00:00
Randy Heit
33a397c04b - Basic Mac support: Everything compiles but does not yet link.
SVN r1780 (trunk)
2009-09-01 02:08:53 +00:00
Randy Heit
732a44b338 - Fixed: M_QuitResponse() tried to play a sound even when none was specified
in the gameinfo.
- Added Yes/No selections for Y/N messages so that you can answer them
  entirely with a joystick.
- Fixed: Starting the menu at the title screen with a key other than Escape
  left the top level menu out of the menu stack.
- Changed the save menu so that cancelling input of a new save name only
  deactivates that control and does not completely close the menus.
- Fixed "any key" messages to override input to menus hidden beneath them and
  to work with joysticks.
- Removed the input parameter from M_StartMessage and the corresponding
  messageNeedsInput global, because it was redundant. Any messages that want
  a Y/N response also supply a callback, and messages that don't care which
  key you press don't supply a callback.
- Changed MKEY_Back so that it cancels out of text entry fields before
  backing to the previous menu, which it already did for the keyboard.
- Changed the menu responder so that key downs always produce results,
  regardless of whether or not an equivalent key is already down.


SVN r1753 (trunk)
2009-08-07 03:30:51 +00:00
Randy Heit
00a6293952 - Fixed: Trying to run with the same FMOD version you compiled with no longer produced
sound. (But compiling with 4.26.xx and using an older DLL was fine.)

SVN r1718 (trunk)
2009-07-15 22:27:38 +00:00
Randy Heit
f74f6a1659 - Split the joystick menu into two parts: A top level with general options
and a list of all attached controllers, and a second level for configuring
  an individual controller.
- Fixed: Pressing Up at the top of a menu with more lines than fit on screen
  would find an incorrect bottom position if the menu had a custom top height.
- Added the cvars joy_dinput, joy_ps2raw, and joy_xinput to enable/disable
  specific game controller input systems independant of each other.
- Device change broadcasts are now sent to the Doom event queue, so
  device scanning can be handled in one common place.
- Added a fast version of IsXInputDevice that uses the Raw Input device
  list, because querying WMI for this information is painfully slow.
- Added support for compiling with FMOD Ex 4.26+ and running the game
  with an older DLL. This combination will now produce sound.


SVN r1717 (trunk)
2009-07-15 05:53:06 +00:00
Randy Heit
ae2c4a6dbc - Relaxed the FMOD version check. Anything from 4.20.00 through 4.26.xx should work
for both building and running now.

SVN r1712 (trunk)
2009-07-06 23:42:00 +00:00
Randy Heit
17418316ea - Added support for FMOD Ex 4.25.01+, which added a geometry parameter to getCPUUsage().
SVN r1710 (trunk)
2009-07-06 03:51:43 +00:00
Randy Heit
a8de4fc2da - Fixed compilation of i_keyboard.cpp with MinGW, because w32api still doesn't have
everything that was new for XP.
- Swapped snes_spc out for the full Game Music Emu library.

SVN r1631 (trunk)
2009-06-03 03:05:02 +00:00
Randy Heit
993fcd233a - Fixed: MinGW needs malloc.h for alloca().
SVN r1529 (trunk)
2009-04-08 04:57:17 +00:00
Christoph Oelckers
a908834a11 - added some default definitions for constants that may miss in some headers.
- replaced __va_copy with va_copy per Chris's suggestion.
- replaced #include <malloc.h> with #include <stdlib.h> where possible.


SVN r1524 (trunk)
2009-04-07 07:06:07 +00:00
Randy Heit
35fe9ae690 - Made fmodex.dll delay-loaded so the game should be runnable on Windows 95
again, though without sound.



SVN r1506 (trunk)
2009-03-26 02:33:37 +00:00
Randy Heit
aee99c2bd0 - Added an SDL output plugin, so FMOD can produce sound using SDL's audio
support instead of its own OSS/ALSA/ESD support. This is selectable by
  setting snd_output to "sdl".



SVN r1473 (trunk)
2009-03-11 04:56:43 +00:00
Randy Heit
47abeec7d9 - Fixed: On Linux systems with ALSA but no OSS support, trying to start
the sound system with snd_output default would fail instead of trying
  to use ALSA.



SVN r1472 (trunk)
2009-03-11 01:12:51 +00:00
Randy Heit
4535382031 - The SFX Reverb unit is now moved so that it serves as an input to the water
effect rather than as an input to the ChannelGroup Target Unit. This means
  the water effect is now applied after any room reverb, rather than in
  parallel with it.
- Fixed: UI sounds should not have reverb applied to them.


SVN r1450 (trunk)
2009-02-26 06:23:05 +00:00
Randy Heit
a18018f607 - Removed the delay for starting all sounds from one tic at exactly the same
DSP position. Without also synchronizing the stopping of sounds, it can
  cause problems with things like the chainsaw where the sound is stopped and
  immediately restarted, causing occasional gaps between the stopping of the
  sound and the starting of the new one. (I added the start synchronization to
  combat flanging of paired moving polyobjects when moving around, but I think
  just removing velocity from the player for sound calculations takes care of
  that well enough.)


SVN r1447 (trunk)
2009-02-25 21:31:14 +00:00
Randy Heit
07a2da40f1 - Changed FMOD init() to allocate some virtual channels.
SVN r1387 (trunk)
2009-02-01 04:10:36 +00:00
Christoph Oelckers
987c878a9a - Fixed: FMOD calls for setting the water reverb must check the return code
for errors.
- Fixed: Hexen's dual attack weapons must check distance to targets in 3D,
  not 2D.


SVN r1336 (trunk)
2008-12-30 23:27:27 +00:00
Randy Heit
f435576f1d - Increased maximum number of per-pattern rows for the XM loader from
256 to 1024 to deal with a module that otherwise would not load.
- Removed the artificial restriction on not supporting Vorbis-compressed
  samples in XMs if they are stereo, since it turns out that OggMod does
  support them.


SVN r1316 (trunk)
2008-12-16 00:59:20 +00:00
Randy Heit
3b4479df44 - Removed GC barriers from the sound channels. If we must, we can always
do it the old way and scan every channel to see if it matches an actor/
  entchannel pair.
- Fixed: S_RelinkSounds() did not move the SoundChans bitfield to the new
  actor.
- Fixed: Stolen channels could be kept around by the high-level channels
  indefinitely.


SVN r1289 (trunk)
2008-11-15 01:01:04 +00:00
Randy Heit
c1cefee2f4 - Made sdl/i_system.cpp:I_GetTimePolled() functionally equivalent to the
Win32 version.
- Updated fmod_wrap.h and fmodsound.cpp for FMOD 4.20.
- GCC warning removal.


SVN r1286 (trunk)
2008-11-09 05:50:54 +00:00
Christoph Oelckers
71dd0d0269 - another #include fix for Linux.
- Added a 'logarithmic' parameter to S_GetRolloff.


SVN r1235 (trunk)
2008-09-18 21:00:12 +00:00
Randy Heit
00b21e4b96 - Apparently, YASM is not a suitable substitute for NASM when doing Win32 builds.
- Removed extraneous printf parameter for Texman.Init startup message.
- Added newlines to the ends of a few headers that were missing them.
- Fixed more GCC errors/warnings.

SVN r1232 (trunk)
2008-09-17 20:24:08 +00:00
Christoph Oelckers
d1f8518a79 - Separated low level sound code from all high level dependencies.
SVN r1228 (trunk)
2008-09-15 18:18:04 +00:00
Christoph Oelckers
5d9483b632 - Moved some more high level sound logic out of FMODSoundRenderer:
The rolloff and channel ended callbacks now call functions in s_sound.cpp
  instead of working on the data itself and GSnd->StopSound has been replaced
  with S_StopChannel.


SVN r1227 (trunk)
2008-09-15 16:02:05 +00:00
Christoph Oelckers
760f70d3f1 - Changed compilation for g_doom, g_heretic, g_hexen and g_strife folders
so that all files are included by a central one instead of compiling 
  each one separately. This speeds up the compilation process by 25%
  when doing a complete rebuild in Visual C.
- Cleaned up more header dependencies.

SVN r1226 (trunk)
2008-09-15 14:11:05 +00:00
Randy Heit
d22f3ebfa6 - Added a check for D3DLINECAPS_ANTIALIAS, but this is complicated by the
fact that NVidia's don't report it, even though they support it. If there
  are any cards that no longer have antialised lines on the automap, please
  let me know.
- Added vid_hwaalines cvar to force antialiased lines off for the
  Direct3D renderer, in case it doesn't really support them.


SVN r1210 (trunk)
2008-09-10 03:33:02 +00:00
Christoph Oelckers
d02097976e - Fixed: The new rolloff values being stored in FSoundChan need to be serialized
for savegames.


SVN r1209 (trunk)
2008-09-09 21:23:09 +00:00
Christoph Oelckers
e033cbf9f8 - Since loading of the sound lump is now done in S_LoadSound I added an IsNull
method to the SoundRenderer class so that this function doesn't need to 
  load the sound for the NullSoundRenderer.
- Took some more non-FMOD related code out of fmodsound.cpp, including the
  code that checks for raw and Doom sounds. This means that sfxinfo_t is no
  longer needed in the SoundRenderer class so I took out all references to it.


SVN r1208 (trunk)
2008-09-09 20:49:53 +00:00