AuroraMiddleware/AuroraOpenALSoft
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity
3,204 Commits 1 Branch 0 Tags 22 MiB
e64d0f83cd
Commit Graph

1312 Commits

Author SHA1 Message Date
Chris Robinson
cd983245f1 Return a sample pointer from resamplers 
Both resampling and filtering now avoid copying samples when they no-op.
 2014-05-19 05:46:01 -07:00
Chris Robinson
8e04a8a022 Put per-channel filter properties together 2014-05-19 02:24:31 -07:00
Chris Robinson
ef9dfe3772 Move an HRTF mixer parameter and shorten a couple variable names 2014-05-18 11:05:38 -07:00
Chris Robinson
9317ec59b9 Don't pass the SendParams to the wet-path mixer 2014-05-18 10:35:11 -07:00
Chris Robinson
c9083d04fa Don't pass the DirectParams to the dry-path mixer 2014-05-18 10:24:07 -07:00
Chris Robinson
5a1abf6918 Use different parameters for HRTF mixers 2014-05-18 09:31:08 -07:00
Chris Robinson
29b5dae6aa Pass some DirectParams as function parameters 2014-05-18 08:59:10 -07:00
Chris Robinson
a5631e05cc Add a ALfilterState method to process multiple samples at once 2014-05-18 06:23:20 -07:00
Chris Robinson
cd53a4b74c Reset filter properties when setting a null filter type 2014-05-17 08:05:49 -07:00
Chris Robinson
190ab90c0d Support the band-pass filter 2014-05-17 08:04:14 -07:00
Chris Robinson
82dd2d875e Apply high-pass source filters as needed 2014-05-17 07:54:25 -07:00
Chris Robinson
3a26ebef0b Implementing handling high-pass filter properties 2014-05-17 07:29:50 -07:00
Chris Robinson
0b5b39d5e6 Add a flag to specify when the low-pass filter needs to apply 2014-05-17 07:17:48 -07:00
Chris Robinson
1efddac3db Document the different filter types, and combine some split lines 2014-05-17 02:09:43 -07:00
Chris Robinson
c5b25ba2a7 Rename freq_scale parameter to freq_mult 2014-05-17 01:32:55 -07:00
Chris Robinson
f0797e27f3 Initialize newformat to AL_NONE instead of a valid format 2014-05-15 01:11:38 -07:00
Chris Robinson
debd5346fb Add a couple asserts to ensure a proper buffer 
It should not be possible for a playing or paused source to not have a
valid buffer, but Clang's static analyzer doesn't know that. Hopefully
an assert will convince it.
 2014-05-14 03:40:01 -07:00
Chris Robinson
1d2504d12e Make RefCount a non-integer type 
It should only be accessed through the appropriate functions to ensure proper
atomicity.
 2014-05-14 02:47:07 -07:00
Chris Robinson
4454ae25c7 Add HF Reference as a filter property 2014-05-14 01:24:18 -07:00
Chris Robinson
3582a44dff Make LOWPASSFREQREF a float value 2014-05-11 10:09:52 -07:00
Chris Robinson
343200d229 Store the filter reference frequency in the source 2014-05-11 10:07:17 -07:00
Chris Robinson
b89cc3417b Avoid locking the mixer when unqueueing buffers 2014-05-11 03:52:22 -07:00
Chris Robinson
9f8615c670 Avoid accessing the source's buffer queue head multiple times 2014-05-11 03:11:35 -07:00
Chris Robinson
851a917b03 Use a struct to store the source's direct gain/gainhf properties 2014-05-11 01:40:44 -07:00
Chris Robinson
ddd1550a8b Avoid locking the mixer while queueing buffers onto a source. 2014-05-10 08:55:28 -07:00
Chris Robinson
e9297d9744 Mark the buffer queue pointers volatile 2014-05-10 08:03:53 -07:00
Chris Robinson
3e28fb33a3 Add a missing pointer init 2014-05-10 07:57:35 -07:00
Chris Robinson
9642bf637d Explicit init the source queue and current_buffer pointers 2014-05-10 06:32:39 -07:00
Chris Robinson
a455704148 Use a RWLock to help protect the source's buffer queue 
In some instances this allows to to remove the device/mixer lock, or reduce how
long it's held.
 2014-05-10 05:07:13 -07:00
Chris Robinson
eebde08e65 Don't explicitly store the BuffersInQueue 2014-05-10 03:33:41 -07:00
Chris Robinson
c4383b65e2 Store the current buffer queue item, rather than played buffer count 2014-05-10 03:21:40 -07:00
Chris Robinson
1c01e94237 Move the static_assert definition to its own header 2014-05-07 02:28:25 -07:00
Chris Robinson
0ea979a262 Move some headers to include/ 
Note, these are not installed. Only headers in include/AL/ are installed.
 2014-05-06 18:29:53 -07:00
Chris Robinson
d066c7b124 Better pack HRTF mixing properties 2014-05-03 18:59:26 -07:00
Chris Robinson
b2e533fbfc Clamp the current and target gain lower bound to epsilon 
Should give a bit more wiggle room for the gain stepping to get lower than the
silence threshold.
 2014-05-03 17:51:06 -07:00
Chris Robinson
149df6d4db Convert the mmdevapi backend to the new backend API 2014-04-25 09:26:53 -07:00
Chris Robinson
32a2f95885 Convert the DSound backend to the new API 2014-04-19 22:54:21 -07:00
Chris Robinson
bb969c3ccc Add a SZFMT macro for a size_t string formatter 2014-04-19 09:42:03 -07:00
Chris Robinson
1d266aa834 Add a GCC-specific STATIC_UPCAST macro that checks the object type 
The check is compile time, and is functionally identical to the old/alternate
version.
 2014-04-19 03:11:23 -07:00
Chris Robinson
59fc9aac0e Use C11 alignas when available 2014-04-19 02:11:04 -07:00
Chris Robinson
c041a99f72 Simplify some error checking 2014-04-18 06:54:18 -07:00
Chris Robinson
b8d56190d1 Check pthread_mutex_unlock for errors 2014-04-18 06:39:46 -07:00
Chris Robinson
1353cb0d88 Rename AL_ONCE_INIT to AL_ONCE_FLAG_INIT 2014-04-18 02:21:58 -07:00
Chris Robinson
579e7c10c2 Check TlsSetValue for error 2014-04-17 22:57:16 -07:00
Chris Robinson
6c8bf9ec42 Rename althread_once to be more C11-like 2014-04-17 21:39:51 -07:00
Chris Robinson
5abefaed0a Use the thread ID for althrd_t on Windows 2014-04-17 21:01:54 -07:00
Chris Robinson
36df67f546 Rename SetThreadName to althrd_setname 2014-04-17 20:41:32 -07:00
Chris Robinson
4b5e8b8c40 Don't inline al_nssleep 2014-04-17 09:14:03 -07:00
Chris Robinson
d1f1a1d056 Make and use a C11-like altimespec_get wrapper function 2014-04-17 09:03:57 -07:00
Chris Robinson
c3b1c31d9b Rename althread_key_ wrappers to altss_ and move it to threads.h/c 2014-04-17 00:11:12 -07:00
Chris Robinson
8cc3d05949 Fix some almtx_ return values 2014-04-16 08:21:45 -07:00
Chris Robinson
2149ccd985 Remove almtx_normal and almtx_errorcheck 2014-04-16 08:00:54 -07:00
Chris Robinson
959d75edc8 Fix althrd_sleep return value 2014-04-16 07:18:28 -07:00
Chris Robinson
b020dd13fd Avoid using a Sleep() wrapper 2014-04-16 06:59:44 -07:00
Chris Robinson
d124aee4d7 Remove the old thread wrappers for the new ones 2014-04-16 06:11:40 -07:00
Chris Robinson
18ab9cbbdd Implement a C11-like thread wrapper and use it in mmdevapi and pulseaudio 2014-04-16 05:19:34 -07:00
Chris Robinson
29cb5058c0 Use a C11-like mutex wrapper instead of CRITICAL_SECTIONs 2014-04-16 01:39:11 -07:00
Chris Robinson
9c70ca9da6 Avoid memory leak if preset ID lookup fails 2014-04-14 01:33:31 -07:00
Chris Robinson
e76bc53d37 Insert all new effect slots into the active effect slots at once 2014-04-10 20:49:01 -07:00
Chris Robinson
c6821e5dd1 Use C11's static_assert when available 2014-04-07 11:48:28 -07:00
Chris Robinson
5e0fc84f49 Make HRTF stepping values per-channel 2014-04-05 10:24:13 -07:00
Chris Robinson
7b93e10f7a Use an al_string for the device name 2014-03-28 22:11:34 -07:00
Chris Robinson
045959e9c0 Use C99 VLA instead of alloca when available 2014-03-25 18:16:03 -07:00
Chris Robinson
03fd2b8266 Remove setState from the MidiSynth vtable 2014-03-23 18:34:49 -07:00
Chris Robinson
de5c54db09 Remove an unused method 2014-03-23 17:47:58 -07:00
Chris Robinson
e2cb41dc61 Remove the last bits of the predictive sample processing 2014-03-23 17:33:57 -07:00
Chris Robinson
72986882f5 Remove the click removal buffers for auxiliary effect slots 2014-03-23 16:28:55 -07:00
Chris Robinson
52deb557d5 Add gain stepping to the send mixers 2014-03-23 16:11:21 -07:00
Chris Robinson
55f851093f Remove the now-unneeded click removal buffers for the device 
They are still there for auxiliary sends. However, they should go away soon
enough too, and then we won't have to mess around with calculating extra
"predictive" samples in the mixer.
 2014-03-23 07:31:33 -07:00
Chris Robinson
81e049bd47 Step mixing gains per-sample for non-HRTF mixing 
This fades the dry mixing gains using a logarithmic curve, which should produce
a smoother transition than a linear one. It functions similarly to a linear
fade except that

step = (target - current) / numsteps;
...
gain += step;

becomes

step = powf(target / current, 1.0f / numsteps);
...
gain *= step;

where 'target' and 'current' are clamped to a lower bound that is greater than
0 (which makes no sense on a logarithmic scale).

Consequently, the non-HRTF direct mixers do not do not feed into the click
removal and pending click buffers, as this per-sample fading would do an
adequate job of stopping clicks and pops caused by extreme gain changes. These
buffers should be removed shortly.
 2014-03-23 06:57:00 -07:00
Chris Robinson
0ce0a88fd6 Move the step counter and moving flag to DirectParams 2014-03-23 03:03:03 -07:00
Chris Robinson
b0d511a860 Store the HrtfState directly in the DirectParams 2014-03-23 02:45:50 -07:00
Chris Robinson
9959972c7e Add a stub 'soft' MIDI synth handler 
Eventually this one will be used to handle MIDI internally, using our own
mixers and resamplers.
 2014-03-22 21:53:00 -07:00
Chris Robinson
842aeb888b Move some HRTF and mixer structs to alu.h 2014-03-22 02:39:57 -07:00
Chris Robinson
0a030c2bd9 Use a void* for the effect state Delete method param 2014-03-21 23:56:18 -07:00
Chris Robinson
8c33b4d742 Increase the vector reserve as needed when pushing in new items 2014-03-21 16:35:38 -07:00
Chris Robinson
ff63188cc2 Add a generic vector interface and use it for the active effect slots 2014-03-21 01:23:01 -07:00
Chris Robinson
e6e3937fa9 Use flexible array members to pad the device and context structs 
This helps avoid the convoluted math otherwise required to ensure the default
slot and listener, respectively, are aligned.
 2014-03-20 14:49:46 -07:00
Chris Robinson
168149ce9d Keep track of the mix count 
The purpose of this is to provide a safe way to be able to "swap" resources
used by the mixer from other threads without the need to block the mixer, as
well as a way to track when mixes have occurred. The idea is two-fold:

It provides a way to safely swap resources. If the mixer were to (atomically)
get a reference to an object to access it from, another thread would be able
allocate and prepare a new object then swap the reference to it with the stored
one. The other thread would then be able to wait until (count&1) is clear,
indicating the mixer is not running, before safely freeing the old object for
the mixer to use the new one.

It also provides a way to tell if the mixer has run. With this, a thread would
be able to read multiple values, which could be altered by the mixer, without
requiring a mixer lock. Comparing the before and after counts for inequality
would signify if the mixer has (started to) run, indicating the values may be
out of sync and should try getting them again. Of course, it will still need
something like a RWLock to ensure another (non-mixer) thread doesn't try to
write to the values at the same time.

Note that because of the possibility of overflow, the counter is not reliable
as an absolute count.
 2014-03-19 19:00:54 -07:00
Chris Robinson
0c5cbafcd8 Use a union to combine HRTF and non-HRTF mixer params 2014-03-19 17:55:09 -07:00
Chris Robinson
6ee54fb1f3 Store some source mixing parameters in the active source struct 2014-03-19 13:14:11 -07:00
Chris Robinson
d6f7aac1bb Use a separate struct for tracking active sources 2014-03-18 19:56:25 -07:00
Chris Robinson
dc74473025 Store the old-style backend funcs in the wrapper 2014-03-17 10:54:27 -07:00
Chris Robinson
bb787ba8ee Avoid GCC's macro arg concat extension with IDE parsing 2014-03-17 09:16:33 -07:00
Chris Robinson
cde2f825d5 Use the correct array size 2014-03-08 22:48:36 -08:00
Chris Robinson
14a04020ff Improve int-to-float and uint-to-float conversions 2014-03-08 22:34:07 -08:00
Chris Robinson
2e279684d7 Only require MSADPCM block alignment to be a multiple of 2 2014-03-06 21:42:03 -08:00
Chris Robinson
7657fbb296 Only sign-expand the nibble when needed 2014-03-06 19:14:19 -08:00
Chris Robinson
143f786d1a Move the sample conversion routines to a separate file 2014-03-05 16:38:02 -08:00
Chris Robinson
bf4ddfef6d Implement a simplistic MSADPCM encoder 
It's not particularly good, but it's better than silence.
 2014-03-05 07:27:22 -08:00
Chris Robinson
61569e09ed Allocate enough temp space for the ADPCM decoders and encoders 2014-03-05 06:18:55 -08:00
Chris Robinson
a4126b66d4 Use specialized methods for converting ALshort to IMA4 and MSADPCM 
As before, to avoid unnecessary direct copies
 2014-03-05 06:13:00 -08:00
Chris Robinson
f82f3f10ab Use specialized methods for converting IMA4 and MSADPCM to ALshort 
Since the decoder methods already convert to ALshort, there's no need to use a
temp buffer that's just going to copy directly to the destination.
 2014-03-05 04:59:24 -08:00
Chris Robinson
15b68fe694 Use maxi to clamp an int to a lower-bound 2014-03-05 04:37:55 -08:00
Chris Robinson
fb1f9aad6e Expand the sign bit on the initial MSADPCM delta value 2014-03-05 00:30:23 -08:00
Chris Robinson
a1c440bd09 Add an extension to support MSADPCM buffer formats 2014-03-04 22:44:30 -08:00
Chris Robinson
f3a87d72b2 Add an extension to alter the block alignment for buffer unpack/pack ops 
This is for unpacking (reading, e.g. alBufferData) and packing (writing, e.g.
alGetBufferSamplesSOFT) operations. The alignments are specified in sample
frames, with 0 meaning the default (65 for IMA4, 1 otherwise). IMA4 alignment
must be a multiple of 8, plus 1 (e.g. alignment = n*8 + 1), otherwise an error
will occur during (un)packing. Chenging the block alignment does not affect
already-loaded sample data, only future unpack/pack operations... so for
example, this is perfectly valid:

// Load mono IMA4 data with a block alignment of 1024 bytes, or 2041 sample
// frames.
alBufferi(buffer, AL_UNPACK_BLOCK_ALIGNMENT_SOFT, 2041);
alBufferData(buffer, AL_FORMAT_MONO_IMA4, data, data_len, srate);
alBufferi(buffer, AL_UNPACK_BLOCK_ALIGNMENT_SOFT, 0);
 2014-03-04 05:53:35 -08:00
Chris Robinson
b0a6b6427c Parameterize the block alignment 2014-03-04 04:29:35 -08:00
Chris Robinson
a6eb38ea7f Use alloca for temp space decoding/encoding IMA4 blocks 2014-03-03 20:02:15 -08:00
Chris Robinson
a2d9133ffc Reduce explicit template declarations and ignore IMA4-to-IMA4 conversions 
Since we never store IMA4 data in buffers, we should only ever convert from
(user input) or to (user output) IMA4. Once we allow user-specified pack/unpack
block alignment, this wouldn't be a simple memcpy anyway.
 2014-03-03 17:25:46 -08:00