3 Fixed CoreAudio capture.
5 Fixed air absorption strength.
7 Fixed handling 5.1 devices on Windows that use Rear channels instead of
10 Fixed some compilation issues on MinGW.
12 Fixed ALSA not being used on some systems without PipeWire and PulseAudio.
14 Fixed OpenSL capturing noise.
16 Fixed Oboe capture failing with some buffer sizes.
18 Added checks for the runtime PipeWire version. The same or newer version
19 than is used for building will be needed at runtime for the backend to
22 Separated 3D7.1 into its own speaker configuration.
26 Implemented the ALC_SOFT_reopen_device extension. This allows for moving
27 devices to different outputs without losing object state.
29 Implemented the ALC_SOFT_output_mode extension.
31 Implemented the AL_SOFT_callback_buffer extension.
33 Implemented the AL_SOFT_UHJ extension. This supports native UHJ buffer
34 formats and Super Stereo processing.
36 Implemented the legacy EAX extensions. Enabled by default only on Windows.
38 Improved sound positioning stability when a source is near the listener.
40 Improved the default 5.1 output decoder.
42 Improved the high frequency response for the HRTF second-order ambisonic
45 Improved SoundIO capture behavior.
47 Fixed UHJ output on NEON-capable CPUs.
49 Fixed redundant effect updates when setting an effect property to the
52 Fixed WASAPI capture using really low sample rates, and sources with very
53 high pitch shifts when using a bsinc resampler.
55 Added a PipeWire backend.
57 Added enumeration for the JACK and CoreAudio backends.
59 Added optional support for RTKit to get real-time priority. Only used as a
60 backup when pthread_setschedparam fails.
62 Added an option for JACK playback to render directly in the real-time
63 processing callback. For lower playback latency, on by default.
65 Added an option for custom JACK devices.
67 Added utilities to encode and decode UHJ audio files. Files are decoded to
68 the .amb format, and are encoded from libsndfile-compatible formats.
70 Added an in-progress extension to hold sources in a playing state when a
71 device disconnects. Allows devices to be reset or reopened and have sources
72 resume from where they left off.
74 Lowered the priority of the JACK backend. To avoid it getting picked when
75 PipeWire is providing JACK compatibility, since the JACK backend is less
76 robust with auto-configuration.
80 Improved alext.h's detection of standard types.
82 Improved slightly the local source position when the listener and source
85 Improved click/pop prevention for sounds that stop prematurely.
87 Fixed compilation for Windows ARM targets with MSVC.
89 Fixed ARM NEON detection on Windows.
91 Fixed CoreAudio capture when the requested sample rate doesn't match the
94 Fixed OpenSL capture desyncing from the internal capture buffer.
96 Fixed sources missing a batch update when applied after quickly restarting
99 Fixed missing source stop events when stopping a paused source.
101 Added capture support to the experimental Oboe backend.
105 Updated library codebase to C++14.
107 Implemented the AL_SOFT_effect_target extension.
109 Implemented the AL_SOFT_events extension.
111 Implemented the ALC_SOFT_loopback_bformat extension.
113 Improved memory use for mixing voices.
115 Improved detection of NEON capabilities.
117 Improved handling of PulseAudio devices that lack manual start control.
119 Improved mixing performance with PulseAudio.
121 Improved high-frequency scaling quality for the HRTF B-Format decoder.
123 Improved makemhr's HRIR delay calculation.
125 Improved WASAPI capture of mono formats with multichannel input.
127 Reimplemented the modulation stage for reverb.
129 Enabled real-time mixing priority by default, for backends that use the
130 setting. It can still be disabled in the config file.
132 Enabled dual-band processing for the built-in quad and 7.1 output decoders.
134 Fixed a potential crash when deleting an effect slot immediately after the
135 last source using it stops.
137 Fixed building with the static runtime on MSVC.
139 Fixed using source stereo angles outside of -pi...+pi.
141 Fixed the buffer processed event count for sources that start with empty
144 Fixed trying to open an unopenable WASAPI device causing all devices to
147 Fixed stale devices when re-enumerating WASAPI devices.
149 Fixed using unicode paths with the log file on Windows.
151 Fixed DirectSound capture reporting bad sample counts or erroring when
154 Added an in-progress extension for a callback-driven buffer type.
156 Added an in-progress extension for higher-order B-Format buffers.
158 Added an in-progress extension for convolution reverb.
160 Added an experimental Oboe backend for Android playback. This requires the
161 Oboe sources at build time, so that it's built as a static library included
164 Added an option for auto-connecting JACK ports.
166 Added greater-than-stereo support to the SoundIO backend.
168 Modified the mixer to be fully asynchronous with the external API, and
169 should now be real-time safe. Although alcRenderSamplesSOFT is not due to
170 locking to check the device handle validity.
172 Modified the UHJ encoder to use an all-pass FIR filter that's less harmful
173 to non-filtered signal phase.
175 Converted examples from SDL_sound to libsndfile. To avoid issues when
176 combining SDL2 and SDL_sound.
178 Worked around a 32-bit GCC/MinGW bug with TLS destructors. See:
179 https://gcc.gnu.org/bugzilla/show_bug.cgi?id=83562
181 Reduced the maximum number of source sends from 16 to 6.
183 Removed the QSA backend. It's been broken for who knows how long.
185 Got rid of the compile-time native-tools targets, using cmake and global
186 initialization instead. This should make cross-compiling less troublesome.
190 Implemented the AL_SOFT_direct_channels_remix extension. This extends
191 AL_DIRECT_CHANNELS_SOFT to optionally remix input channels that don't have
192 a matching output channel.
194 Implemented the AL_SOFT_bformat_ex extension. This extends B-Format buffer
195 support for N3D or SN3D scaling, or ACN channel ordering.
197 Fixed a potential voice leak when a source is started and stopped or
198 restarted in quick succession.
200 Fixed a potential device reset failure with JACK.
202 Improved handling of unsupported channel configurations with WASAPI. Such
203 setups will now try to output at least a stereo mix.
205 Improved clarity a bit for the HRTF second-order ambisonic decoder.
207 Improved detection of compatible layouts for SOFA files in makemhr and
210 Added the ability to resample HRTFs on load. MHR files no longer need to
211 match the device sample rate to be usable.
213 Added an option to limit the HRTF's filter length.
217 Converted the library codebase to C++11. A lot of hacks and custom
218 structures have been replaced with standard or cleaner implementations.
220 Partially implemented the Vocal Morpher effect.
222 Fixed the bsinc SSE resamplers on non-GCC compilers.
224 Fixed OpenSL capture.
226 Fixed support for extended capture formats with OpenSL.
228 Fixed handling of WASAPI not reporting a default device.
230 Fixed performance problems relating to semaphores on macOS.
232 Modified the bsinc12 resampler's transition band to better avoid aliasing
235 Modified alcResetDeviceSOFT to attempt recovery of disconnected devices.
237 Modified the virtual speaker layout for HRTF B-Format decoding.
239 Modified the PulseAudio backend to use a custom processing loop.
241 Renamed the makehrtf utility to makemhr.
243 Improved the efficiency of the bsinc resamplers when up-sampling.
245 Improved the quality of the bsinc resamplers slightly.
247 Improved the efficiency of the HRTF filters.
249 Improved the HRTF B-Format decoder coefficient generation.
251 Improved reverb feedback fading to be more consistent with pan fading.
253 Improved handling of sources that end prematurely, avoiding loud clicks.
255 Improved the performance of some reverb processing loops.
257 Added fast_bsinc12 and 24 resamplers that improve efficiency at the cost of
258 some quality. Notably, down-sampling has less smooth pitch ramping.
260 Added support for SOFA input files with makemhr.
262 Added a build option to use pre-built native tools. For cross-compiling,
263 use with caution and ensure the native tools' binaries are kept up-to-date.
265 Added an adjust-latency config option for the PulseAudio backend.
267 Added basic support for multi-field HRTFs.
269 Added an option for mixing first- or second-order B-Format with HRTF
270 output. This can improve HRTF performance given a number of sources.
272 Added an RC file for proper DLL version information.
274 Disabled some old KDE workarounds by default. Specifically, PulseAudio
275 streams can now be moved (KDE may try to move them after opening).
279 Implemented capture support for the SoundIO backend.
281 Fixed source buffer queues potentially not playing properly when a queue
284 Fixed possible unexpected failures when generating auxiliary effect slots.
286 Fixed a crash with certain reverb or device settings.
288 Fixed OpenSL capture.
290 Improved output limiter response, better ensuring the sample amplitude is
295 Implemented the ALC_SOFT_device_clock extension.
297 Implemented the Pitch Shifter, Frequency Shifter, and Autowah effects.
299 Fixed compiling on FreeBSD systems that use freebsd-lib 9.1.
301 Fixed compiling on NetBSD.
303 Fixed the reverb effect's density scale and panning parameters.
305 Fixed use of the WASAPI backend with certain games, which caused odd COM
306 initialization errors.
308 Increased the number of virtual channels for decoding Ambisonics to HRTF
311 Changed 32-bit x86 builds to use SSE2 math by default for performance.
312 Build-time options are available to use just SSE1 or x87 instead.
314 Replaced the 4-point Sinc resampler with a more efficient cubic resampler.
316 Renamed the MMDevAPI backend to WASAPI.
318 Added support for 24-bit, dual-ear HRTF data sets. The built-in data set
319 has been updated to 24-bit.
321 Added a 24- to 48-point band-limited Sinc resampler.
323 Added an SDL2 playback backend. Disabled by default to avoid a dependency
326 Improved the performance and quality of the Chorus and Flanger effects.
328 Improved the efficiency of the band-limited Sinc resampler.
330 Improved the Sinc resampler's transition band to avoid over-attenuating
333 Improved the performance of some filter operations.
335 Improved the efficiency of object ID lookups.
337 Improved the efficienty of internal voice/source synchronization.
339 Improved AL call error logging with contextualized messages.
341 Removed the reverb effect's modulation stage. Due to the lack of reference
342 for its intended behavior and strength.
346 Fixed resetting the FPU rounding mode after certain function calls on
349 Fixed use of SSE intrinsics when building with Clang on Windows.
351 Fixed a crash with the JACK backend when using JACK1.
353 Fixed use of pthread_setnane_np on NetBSD.
355 Fixed building on FreeBSD with an older freebsd-lib.
357 OSS now links with libossaudio if found at build time (for NetBSD).
361 Fixed an issue where resuming a source might not restart playing it.
363 Fixed PulseAudio playback when the configured stream length is much less
364 than the requested length.
366 Fixed MMDevAPI capture with sample rates not matching the backing device.
368 Fixed int32 output for the Wave Writer.
370 Fixed enumeration of OSS devices that are missing device files.
372 Added correct retrieval of the executable's path on FreeBSD.
374 Added a config option to specify the dithering depth.
376 Added a 5.1 decoder preset that excludes front-center output.
380 Implemented the AL_EXT_STEREO_ANGLES and AL_EXT_SOURCE_RADIUS extensions.
382 Implemented the AL_SOFT_gain_clamp_ex, AL_SOFT_source_resampler,
383 AL_SOFT_source_spatialize, and ALC_SOFT_output_limiter extensions.
385 Implemented 3D processing for some effects. Currently implemented for
386 Reverb, Compressor, Equalizer, and Ring Modulator.
388 Implemented 2-channel UHJ output encoding. This needs to be enabled with a
389 config option to be used.
391 Implemented dual-band processing for high-quality ambisonic decoding.
393 Implemented distance-compensation for surround sound output.
395 Implemented near-field emulation and compensation with ambisonic rendering.
396 Currently only applies when using the high-quality ambisonic decoder or
397 ambisonic output, with appropriate config options.
399 Implemented an output limiter to reduce the amount of distortion from
402 Implemented dithering for 8-bit and 16-bit output.
404 Implemented a config option to select a preferred HRTF.
406 Implemented a run-time check for NEON extensions using /proc/cpuinfo.
408 Implemented experimental capture support for the OpenSL backend.
410 Fixed building on compilers with NEON support but don't default to having
413 Fixed support for JACK on Windows.
415 Fixed starting a source while alcSuspendContext is in effect.
417 Fixed detection of headsets as headphones, with MMDevAPI.
419 Added support for AmbDec config files, for custom ambisonic decoder
420 configurations. Version 3 files only.
422 Added backend-specific options to alsoft-config.
424 Added first-, second-, and third-order ambisonic output formats. Currently
425 only works with backends that don't rely on channel labels, like JACK,
428 Added a build option to embed the default HRTFs into the lib.
430 Added AmbDec presets to enable high-quality ambisonic decoding.
432 Added an AmbDec preset for 3D7.1 speaker setups.
434 Added documentation regarding Ambisonics, 3D7.1, AmbDec config files, and
435 the provided ambdec presets.
437 Added the ability for MMDevAPI to open devices given a Device ID or GUID
440 Added an option to the example apps to open a specific device.
442 Increased the maximum auxiliary send limit to 16 (up from 4). Requires
443 requesting them with the ALC_MAX_AUXILIARY_SENDS context creation
446 Increased the default auxiliary effect slot count to 64 (up from 4).
448 Reduced the default period count to 3 (down from 4).
450 Slightly improved automatic naming for enumerated HRTFs.
452 Improved B-Format decoding with HRTF output.
454 Improved internal property handling for better batching behavior.
456 Improved performance of certain filter uses.
458 Removed support for the AL_SOFT_buffer_samples and AL_SOFT_buffer_sub_data
459 extensions. Due to conflicts with AL_EXT_SOURCE_RADIUS.
463 Implemented device enumeration for OSSv4.
465 Fixed building on OSX.
467 Fixed building on non-Windows systems without POSIX-2008.
469 Fixed Dedicated Dialog and Dedicated LFE effect output.
471 Added a build option to override the share install dir.
473 Added a build option to static-link libgcc for MinGW.
477 Fixed building with JACK and without PulseAudio.
479 Fixed building on FreeBSD.
481 Fixed the ALSA backend's allow-resampler option.
483 Fixed handling of inexact ALSA period counts.
485 Altered device naming scheme on Windows backends to better match other
488 Updated the CoreAudio backend to use the AudioComponent API. This clears up
489 deprecation warnings for OSX 10.11, although requires OSX 10.6 or newer.
493 Implemented a JACK playback backend.
495 Implemented the AL_EXT_BFORMAT and AL_EXT_MULAW_BFORMAT extensions.
497 Implemented the ALC_SOFT_HRTF extension.
499 Implemented C, SSE3, and SSE4.1 based 4- and 8-point Sinc resamplers.
501 Implemented a C and SSE based band-limited Sinc resampler. This does 12- to
502 24-point Sinc resampling, and performs anti-aliasing.
504 Implemented B-Format output support for the wave file writer. This creates
505 FuMa-style first-order Ambisonics wave files (AMB format).
507 Implemented a stereo-mode config option for treating stereo modes as either
508 speakers or headphones.
510 Implemented per-device configuration options.
512 Fixed handling of PulseAudio and MMDevAPI devices that have identical
515 Fixed a potential lockup when stopping playback of suspended PulseAudio devices.
517 Fixed logging of Unicode characters on Windows.
519 Fixed 5.1 surround sound channels. By default it will now use the side
520 channels for the surround output. A configuration using rear channels is
523 Fixed the QSA backend potentially altering the capture format.
525 Fixed detecting MMDevAPI's default device.
527 Fixed returning the default capture device name.
529 Fixed mixing property calculations when deferring context updates.
531 Altered the behavior of alcSuspendContext and alcProcessContext to better
532 match certain Windows drivers.
534 Altered the panning algorithm, utilizing Ambisonics for better side and
535 back positioning cues with surround sound output.
537 Improved support for certain older Windows apps.
539 Improved the alffplay example to support surround sound streams.
541 Improved support for building as a sub-project.
543 Added an HRTF playback example.
545 Added a tone generator output test.
547 Added a toolchain to help with cross-compiling to Android.
551 Implemented EFX Chorus, Flanger, Distortion, Equalizer, and Compressor
554 Implemented high-pass and band-pass EFX filters.
556 Implemented the high-pass filter for the EAXReverb effect.
558 Implemented SSE2 and SSE4.1 linear resamplers.
560 Implemented Neon-enhanced non-HRTF mixers.
562 Implemented a QSA backend, for QNX.
564 Implemented the ALC_SOFT_pause_device, AL_SOFT_deferred_updates,
565 AL_SOFT_block_alignment, AL_SOFT_MSADPCM, and AL_SOFT_source_length
568 Fixed resetting mmdevapi backend devices.
570 Fixed clamping when converting 32-bit float samples to integer.
572 Fixed modulation range in the Modulator effect.
574 Several fixes for the OpenSL playback backend.
576 Fixed device specifier names that have Unicode characters on Windows.
578 Added support for filenames and paths with Unicode (UTF-8) characters on
581 Added support for alsoft.conf config files found in XDG Base Directory
582 Specification locations (XDG_CONFIG_DIRS and XDG_CONFIG_HOME, or their
583 defaults) on non-Windows systems.
585 Added a GUI configuration utility (requires Qt 4.8).
587 Added support for environment variable expansion in config options (not
588 keys or section names).
590 Added an example that uses SDL2 and ffmpeg.
592 Modified examples to use SDL_sound.
594 Modified CMake config option names for better sorting.
596 HRTF data sets specified in the hrtf_tables config option may now be
597 relative or absolute filenames.
599 Made the default HRTF data set an external file, and added a data set for
600 48khz playback in addition to 44.1khz.
602 Added support for C11 atomic methods.
604 Improved support for some non-GNU build systems.
608 Fixed a regression with retrieving the source's AL_GAIN property.
612 Fixed device enumeration with the OSS backend.
614 Reorganized internal mixing logic, so unneeded steps can potentially be
615 skipped for better performance.
617 Removed the lookup table for calculating the mixing pans. The panning is
618 now calculated directly for better precision.
620 Improved the panning of stereo source channels when using stereo output.
622 Improved source filter quality on send paths.
624 Added a config option to allow PulseAudio to move streams between devices.
626 The PulseAudio backend will now attempt to spawn a server by default.
628 Added a workaround for a DirectSound bug relating to float32 output.
630 Added SSE-based mixers, for HRTF and non-HRTF mixing.
632 Added support for the new AL_SOFT_source_latency extension.
634 Improved ALSA capture by avoiding an extra buffer when using sizes
635 supported by the underlying device.
637 Improved the makehrtf utility to support new options and input formats.
639 Modified the CFLAGS declared in the pkg-config file so the "AL/" portion of
640 the header includes can optionally be omitted.
642 Added a couple example code programs to show how to apply reverb, and
645 The configuration sample is now installed into the share/openal/ directory
646 instead of /etc/openal.
648 The configuration sample now gets installed by default.