aboutsummaryrefslogtreecommitdiff
path: root/sound
AgeCommit message (Collapse)Author
2003-07-30the VC warning for unknown pragmas is now gloabally disabled, no need to ↵Max Horn
work around it locally svn-id: r9300
2003-07-30Partial fix for bug #780167 ("SIMON1/2: Sound problems"). It seems to fixTorbjörn Andersson
the crash, but the sounds play at the wrong rate, I believe. svn-id: r9298
2003-07-30Fixed MSVC errorsJamieson Christian
svn-id: r9288
2003-07-29removed some debug outputMax Horn
svn-id: r9282
2003-07-29cleanupMax Horn
svn-id: r9281
2003-07-29account for the data in the _buffer, tooMax Horn
svn-id: r9276
2003-07-29after looking at SAM and DOTT, it seems the actor-voice sync actually is ↵Max Horn
*worse* with a non-zero _silenceCut value - setting it to 0 gives perfect sync for me... odd, when did this change? Anyway, removing _silenceCut. If you notice sync problems, tell me svn-id: r9275
2003-07-29some cleanup; moved around some stuff, preparing to unify more code of ↵Max Horn
ChannelMP3CDMusic and ChannelMP3 svn-id: r9274
2003-07-29return immediately after commiting suicide; enable new rate conversion codeMax Horn
svn-id: r9273
2003-07-29hooked up ChannelVorbis to the new rate converter system, using ↵Max Horn
VorbisInputStream svn-id: r9272
2003-07-29added VorbisInputStreamMax Horn
svn-id: r9271
2003-07-29added makeRateConverter factory functionMax Horn
svn-id: r9270
2003-07-29Fix mingw compileTravis Howell
svn-id: r9269
2003-07-29cleanupMax Horn
svn-id: r9266
2003-07-28experimental new rate conversion code (use SOX_HACK to en-/disable it); note ↵Max Horn
that right now only the linear filter works, the high quality resample in resample.cpp is in flux; right now the top priority for me is to get MP3/Vorbis resampling working svn-id: r9261
2003-07-28fix for bug #776169Max Horn
svn-id: r9259
2003-07-28make wrapped stream work (I can hear some music in CoMI now before it ↵Max Horn
segfaults :-) svn-id: r9258
2003-07-28moved st_gcd to resample.cppMax Horn
svn-id: r9257
2003-07-28rewrote the input streams, this should be more straightforward and inline ↵Max Horn
slightly better (the fact that the main AudioInputStream methods are virtual still means that no inlining can take place there; but of course, if we did change that, all the rate converters would have to exist in 8, instead of currently 2, compiled versions) svn-id: r9256
2003-07-28more work on WrappedMemoryStream (not yet tested)Max Horn
svn-id: r9253
2003-07-28instead of 'int channels', use 'bool stereo' (less extensible, but then I ↵Max Horn
don't think we'll ever support 5.1 sound :-)); fixed a bug in st_rate_flow where it sometimes would overflow the output buffer; made CopyRateConverter a template, too, increasing efficency svn-id: r9239
2003-07-28added CopyRateConverter (used if inrate == outrate)Max Horn
svn-id: r9214
2003-07-28use a single converter for both channels (if input data is stereo), for ↵Max Horn
improved efficency; renamed MemoryAudioInputStream -> LinearMemoryStream and use some template voodoo to make the baseclass of Input8bitSignedStream etc. variable; added (commented out) draf of WrappedMemoryStream svn-id: r9209
2003-07-26For for Bug [775654] DOTT: Hanging notesJamieson Christian
Corrected active notes tracking when switching between Type 2 SMF tracks. Although Note Off events from the previous track are properly simulated, those sustaining notes are cleared from the active notes list so that jumps within the new track will not attempt to sustain them again. svn-id: r9194
2003-07-26oops accidentally broke the 'old' 8bit resampling code :-(Max Horn
svn-id: r9192
2003-07-25move AudioInputStream implementations & makeInputStream into separate .cpp fileMax Horn
svn-id: r9190
2003-07-25reduced source code duplicationMax Horn
svn-id: r9189
2003-07-25lots of mixer changes: replaced _volumeTable by _globalVolume (applying ↵Max Horn
volume after resampling is more accurate); made more member vars of SoundMixer protected (and thus added some new getter methods); added (untested) support for a second (stereo) channel when playing MP3 svn-id: r9184
2003-07-25round to nearestMax Horn
svn-id: r9183
2003-07-24whitespaceMax Horn
svn-id: r9181
2003-07-24added volume support; some cleanup; provided exactly as many samples as were ↵Max Horn
requested in the 'resample' algorithm (if possible) svn-id: r9180
2003-07-24added RateConverter classes as thin wrappers around st_* functions; renamed ↵Max Horn
InputStream -> AudioInputStream and cleaned up the hierarchy of its subclasses; added makeInputStream() auxillary function svn-id: r9179
2003-07-24new files, based on SoX (http://sox.sf.net): better resampling code. Note ↵Max Horn
that my mixer.cpp changes are on purpose not yet in CVS since they are not complete. Only reasons I checkin these files is that it's much more comfortable to have CVS, since I need to rewrite parts of resample.cpp now (I already have lots of modifications in). Also expect more OO in the future svn-id: r9176
2003-07-23Fixed regression in clamped_add_16() that kept it from clamping. This wasTorbjörn Andersson
quite noticeable in the BASS floppy intro. Of course, if it turns out that 'int' is also only 16 bits it still won't work, but in that case we run into problems even before this function is called. svn-id: r9145
2003-07-17simplified codeMax Horn
svn-id: r9061
2003-07-14oopsMax Horn
svn-id: r9023
2003-07-14removed whitespaces at line ends; added stopHandleMax Horn
svn-id: r9022
2003-07-12Fixed RTTI warnings with PalmOS portChris Apers
svn-id: r8951
2003-07-12verify chan != 0 even on WinCEMax Horn
svn-id: r8940
2003-07-12Give sample rate to channel classes (WinCE downsampling) / Do not use ↵Nicolas Bacca
dynamic cast with EVC 3 / Fix Tremor Ogg support svn-id: r8937
2003-07-11Reverted to "old" YM3812 (FMOPL) emulator code.Jamieson Christian
Until specific information regarding the benefits of migrating to the new emulator can be made available, the "old" code will remain in effect for the benefit of slower hardware platforms such as some WinCE-based devices. svn-id: r8903
2003-07-10Fix for Bug [766426]: V5 Games: Adlib SFX not loopedJamieson Christian
Modified Smart Jump logic to deal with active notes whose Note On and Note Off events BOTH occur OUTSIDE the range of the jump. While this is not a thorough way to deal with Note On events that occur outside jump points, it at least deals with the issue of long, unchanging Adlib SFX used by some earlier SCUMM games. svn-id: r8892
2003-07-07#include scummsys.h so that our custome free() gets usedMax Horn
svn-id: r8849
2003-07-06mix() is really a private method, only to be called by the audio thread; ↵Max Horn
cover setupPremix by the mutex, too (that helps preventing race conditions upon exit) svn-id: r8809
2003-07-06now that we mutex-protect everything properly, we can get rid of ↵Max Horn
_toBeDestroyed; also, instead of keeping a global _handles array in the mixer, let each Channel manage its own handle svn-id: r8808
2003-07-06the mutex must be locked by everything which might access _channels while ↵Max Horn
the mixer thread is running, because the mixer thread may modify _channels svn-id: r8805
2003-07-06More fixes from HibernatusTravis Howell
svn-id: r8792
2003-07-06cleanup to the mixer helper proc interface and to ChannelStreamMax Horn
svn-id: r8785
2003-07-05protect calls to insertChannel with mutex, too (and switch to using StackLock)Max Horn
svn-id: r8782
2003-07-05updated code to use type MutexRef; added class StackLock (from sdl backend, ↵Max Horn
now in util.*) svn-id: r8777