opus.git
6 years agoRemoves useless VARDECL() of "out"
Jean-Marc Valin [Tue, 1 Oct 2013 23:46:48 +0000 (19:46 -0400)]
Removes useless VARDECL() of "out"

6 years agoReplaces log2() with celt_log2() and fixes a potential divide-by-zero
Jean-Marc Valin [Tue, 1 Oct 2013 23:27:30 +0000 (19:27 -0400)]
Replaces log2() with celt_log2() and fixes a potential divide-by-zero

6 years agoFixes scaling of downmix_float() for fixed-point.
Jean-Marc Valin [Tue, 1 Oct 2013 23:25:40 +0000 (19:25 -0400)]
Fixes scaling of downmix_float() for fixed-point.

The previous version simply produced zeros for fixed-point.

6 years agoFixes missing VARDECL()
Jean-Marc Valin [Tue, 1 Oct 2013 21:16:33 +0000 (17:16 -0400)]
Fixes missing VARDECL()

6 years agoDon't use --dirty in genversion.bat.
Ralph Giles [Tue, 1 Oct 2013 21:07:17 +0000 (14:07 -0700)]
Don't use --dirty in genversion.bat.

This script updates version.mk, which is under version control,
so the tree is always dirty. Until the script is updated to write
package_version instead, incorporating version.mk as a fallback
the way update_version does, it's better not to use --dirty.

6 years agoMerge branch 'exp_surround1'
Jean-Marc Valin [Tue, 1 Oct 2013 21:06:09 +0000 (17:06 -0400)]
Merge branch 'exp_surround1'

6 years agoPass --always to git describe.
Ralph Giles [Tue, 1 Oct 2013 20:05:33 +0000 (13:05 -0700)]
Pass --always to git describe.

This generates a useful version string even for --depth=1 checkouts.
The option was added in git v1.6.6.

6 years agoAvoid using C++ keyword 'new'
Stefan Hacker [Fri, 20 Sep 2013 02:07:53 +0000 (04:07 +0200)]
Avoid using C++ keyword 'new'

Signed-off-by: Jean-Marc Valin <jmvalin@jmvalin.ca>
6 years agoAvoids unnecessary copying in opus_encode for delayed decision exp_surround1
Jean-Marc Valin [Sun, 29 Sep 2013 03:51:11 +0000 (23:51 -0400)]
Avoids unnecessary copying in opus_encode for delayed decision

6 years agoMakes analysis work when encoding more than 20 ms at a time
Jean-Marc Valin [Sat, 28 Sep 2013 23:29:23 +0000 (19:29 -0400)]
Makes analysis work when encoding more than 20 ms at a time

6 years agorun_analysis() doesn't need to return the frame size anymore
Jean-Marc Valin [Sat, 28 Sep 2013 21:22:41 +0000 (17:22 -0400)]
run_analysis() doesn't need to return the frame size anymore

6 years agoRemove trailing whitespace from the license headers.
Ralph Giles [Mon, 16 Sep 2013 21:40:04 +0000 (14:40 -0700)]
Remove trailing whitespace from the license headers.

6 years agoRemove trailing whitespace.
Ralph Giles [Mon, 16 Sep 2013 21:29:24 +0000 (14:29 -0700)]
Remove trailing whitespace.

6 years agoMakes the fixed-point build only run the float analysis at complexity 10
Jean-Marc Valin [Mon, 16 Sep 2013 18:36:53 +0000 (14:36 -0400)]
Makes the fixed-point build only run the float analysis at complexity 10

Default complexity is now 9.

6 years agoMoves frame size selection back out of opus_encode_native()
Jean-Marc Valin [Mon, 16 Sep 2013 17:08:52 +0000 (13:08 -0400)]
Moves frame size selection back out of opus_encode_native()

Should fix delayed decision on surround, but not well tested.

6 years agoVery basic surround rate calibration
Jean-Marc Valin [Sat, 14 Sep 2013 19:46:09 +0000 (15:46 -0400)]
Very basic surround rate calibration

6 years agoAllowing surround to use lower bandwidth than fullband
Jean-Marc Valin [Fri, 13 Sep 2013 21:16:38 +0000 (17:16 -0400)]
Allowing surround to use lower bandwidth than fullband

6 years agoMore surround masking tuning
Jean-Marc Valin [Fri, 13 Sep 2013 19:05:50 +0000 (15:05 -0400)]
More surround masking tuning

6 years agoTuning the surround masking
Jean-Marc Valin [Thu, 12 Sep 2013 07:05:43 +0000 (03:05 -0400)]
Tuning the surround masking

We're now a bit more conservative and only take into account the bands
that are being codec. Also weighting the rate saving by the band width.

6 years agoSurround analysis should now work for non-20ms frame sizes.
Jean-Marc Valin [Thu, 12 Sep 2013 03:34:51 +0000 (23:34 -0400)]
Surround analysis should now work for non-20ms frame sizes.

Not well tested.

6 years agoCode for handling upsampling in surround_analysis()
Jean-Marc Valin [Wed, 11 Sep 2013 04:51:22 +0000 (00:51 -0400)]
Code for handling upsampling in surround_analysis()

Still untested.

6 years agoAdds surround masking to SILK
Jean-Marc Valin [Tue, 10 Sep 2013 05:15:19 +0000 (01:15 -0400)]
Adds surround masking to SILK

6 years agoFixes delayed decision for fixed-point
Jean-Marc Valin [Mon, 9 Sep 2013 21:11:25 +0000 (17:11 -0400)]
Fixes delayed decision for fixed-point

6 years agoAnalysis scaling fixes
Jean-Marc Valin [Mon, 9 Sep 2013 20:39:19 +0000 (16:39 -0400)]
Analysis scaling fixes

This should make the scaling the same for fixed and float. It changes the float
scaling too by normalizing by the number of channels, which matters for
bandwidth detection.

6 years agoMakes speech/music detection work with FIXED_POINT (code still float)
Jean-Marc Valin [Fri, 6 Sep 2013 20:32:50 +0000 (16:32 -0400)]
Makes speech/music detection work with FIXED_POINT (code still float)

6 years agoFirst pass at making the analysis code run with FIXED_POINT
Jean-Marc Valin [Fri, 6 Sep 2013 20:00:39 +0000 (16:00 -0400)]
First pass at making the analysis code run with FIXED_POINT

Code is still float, but at least tonality esitmation seems to work.
Speech/music analysis is still disabled.

6 years agoMoves the analysis back to opus_encode_native() to it runs for all streams.
Jean-Marc Valin [Fri, 6 Sep 2013 01:34:43 +0000 (21:34 -0400)]
Moves the analysis back to opus_encode_native() to it runs for all streams.

I checked that it's not *completely* busted, but it probably needs more testing.

6 years agoAdds trim and dynalloc control to surround masking
Jean-Marc Valin [Thu, 5 Sep 2013 16:49:55 +0000 (12:49 -0400)]
Adds trim and dynalloc control to surround masking

6 years agoMakes surround_analysis() work in fixed-point
Jean-Marc Valin [Sat, 31 Aug 2013 06:05:32 +0000 (02:05 -0400)]
Makes surround_analysis() work in fixed-point

6 years agoSurround masking rewrite
Jean-Marc Valin [Sat, 31 Aug 2013 01:58:02 +0000 (21:58 -0400)]
Surround masking rewrite

The channel energy is now computed using an MDCT in the multi-stream
encoder rather than computing the energy of the stereo downmix.

6 years agoImplements fixed-point silk_LPC_analysis_filter() in terms of celt_fir()
Jean-Marc Valin [Thu, 29 Aug 2013 19:29:02 +0000 (15:29 -0400)]
Implements fixed-point silk_LPC_analysis_filter() in terms of celt_fir()

Saves 2.5% on ARM without any asm. The float build still uses the old
code because celt_fir() then becomes a float function.

6 years agoConverts the fixed-point silk_burg_modified() to use celt_pitch_xcorr()
Jean-Marc Valin [Thu, 29 Aug 2013 01:38:15 +0000 (21:38 -0400)]
Converts the fixed-point silk_burg_modified() to use celt_pitch_xcorr()

bit-exact with original version

6 years agoShare auto-correlation code between SILK and CELT
Jean-Marc Valin [Wed, 28 Aug 2013 21:55:34 +0000 (17:55 -0400)]
Share auto-correlation code between SILK and CELT

6 years agoMakes the SILK pitch search use celt_pitch_xcorr()
Jean-Marc Valin [Mon, 26 Aug 2013 22:54:39 +0000 (18:54 -0400)]
Makes the SILK pitch search use celt_pitch_xcorr()

Should gives us ARM/SSE optimizations for free.

6 years agoFix more VS2010 header paths.
Ralph Giles [Wed, 21 Aug 2013 23:15:57 +0000 (16:15 -0700)]
Fix more VS2010 header paths.

6 years agoFix relative paths the header file in the VS2010 projects.
Ralph Giles [Wed, 21 Aug 2013 23:06:30 +0000 (16:06 -0700)]
Fix relative paths the header file in the VS2010 projects.

This prevented opening the files from the 'Solution Explorer'
interface where the source listing is displayed. Thanks to
Koen Vos for reporting the issue.

6 years agoFixes a denorm issue in the VoIP mode high-pass filter
Jean-Marc Valin [Mon, 12 Aug 2013 18:47:14 +0000 (14:47 -0400)]
Fixes a denorm issue in the VoIP mode high-pass filter

6 years agoUpdate the RTP draft for resubmit
Jean-Marc Valin [Fri, 2 Aug 2013 16:04:50 +0000 (12:04 -0400)]
Update the RTP draft for resubmit

6 years agoFix broken URL in entdec.c.
Timothy B. Terriberry [Mon, 5 Aug 2013 05:47:07 +0000 (22:47 -0700)]
Fix broken URL in entdec.c.

Thanks to thinqa for the report.
Fixes #1968.

6 years agoBetter handling of the dynalloc cap for CBR/CVBR
Jean-Marc Valin [Tue, 30 Jul 2013 09:12:46 +0000 (05:12 -0400)]
Better handling of the dynalloc cap for CBR/CVBR

6 years agoFixes an integer wrap-around due to an exceedingly large QCONST16
Jean-Marc Valin [Thu, 25 Jul 2013 17:52:42 +0000 (13:52 -0400)]
Fixes an integer wrap-around due to an exceedingly large QCONST16

6 years agoAdd SET_MAX_BANDWIDTH to multistream encoder ctl list
Daniel Jensen [Mon, 22 Jul 2013 22:31:31 +0000 (16:31 -0600)]
Add SET_MAX_BANDWIDTH to multistream encoder ctl list

Signed-off-by: Gregory Maxwell <greg@xiph.org>
6 years agoDisable MSVC posix security warnings in the project file.
Ralph Giles [Mon, 15 Jul 2013 23:17:02 +0000 (16:17 -0700)]
Disable MSVC posix security warnings in the project file.

By default Visual Studio warns about various C stdlib calls,
recommending non-portable replacements instead. We disable
this warning in a number of places in the source.

Since they're specific to the Visual Studio build, it's better
to disable them just in the project files where they always
apply to the correct toolchain.

I have only added the disable setting to project files which
need it currently: opus, test_opus_encode, and test_opus_decode.

6 years agoConditionalize #pragma warn for MSVC.
Ralph Giles [Mon, 15 Jul 2013 23:04:14 +0000 (16:04 -0700)]
Conditionalize #pragma warn for MSVC.

This warning disable was conditional on _WIN32, which mingw
also defines, resulting in a warning about the unknown pragma
on gcc. Instead make it depend on something only the Visual
Studio compiler defines.

6 years agoRevert "Fixes a number of double promotions and missing casts."
Gregory Maxwell [Mon, 15 Jul 2013 22:51:24 +0000 (15:51 -0700)]
Revert "Fixes a number of double promotions and missing casts."

This reverts commit 43279728cd39947f95dbf77557a0a698f164fdd1.

6 years agoRevert "Additional MSVC warning missed in the last commit."
Gregory Maxwell [Mon, 15 Jul 2013 22:51:14 +0000 (15:51 -0700)]
Revert "Additional MSVC warning missed in the last commit."

This reverts commit 715b37f6d65b855c2ac2677ecafef6f9f7290de1.

6 years agoAdditional MSVC warning missed in the last commit.
Gregory Maxwell [Mon, 15 Jul 2013 22:31:44 +0000 (15:31 -0700)]
Additional MSVC warning missed in the last commit.

6 years agoFixes a number of double promotions and missing casts.
Gregory Maxwell [Mon, 15 Jul 2013 22:23:03 +0000 (15:23 -0700)]
Fixes a number of double promotions and missing casts.

At a minimum MSVC warns on some of this stuff.

6 years agoAlways try to update the version when autogen.sh is run
Ron [Sat, 13 Jul 2013 07:53:04 +0000 (17:23 +0930)]
Always try to update the version when autogen.sh is run

This avoids at least one case where ./autogen.sh && ./configure && make
will re-run configure because the makefile rules updated something that
it depends upon. Pulling a new version from git will change the version
so we should update that at the first step rather than iterating after
the last one.

6 years agoUpdate draft for the Opus RFC
Jean-Marc Valin [Sat, 13 Jul 2013 03:52:00 +0000 (23:52 -0400)]
Update draft for the Opus RFC

6 years agoFixes MSVC warnings
Jean-Marc Valin [Sat, 13 Jul 2013 02:59:29 +0000 (22:59 -0400)]
Fixes MSVC warnings

6 years agoUpdates version.mk for 1.1-beta until we can get rid of it v1.1-beta
Jean-Marc Valin [Fri, 12 Jul 2013 14:08:52 +0000 (10:08 -0400)]
Updates version.mk for 1.1-beta until we can get rid of it

6 years agoFixes a denorm problem when the input goes silent after active audio
Jean-Marc Valin [Fri, 12 Jul 2013 05:22:09 +0000 (01:22 -0400)]
Fixes a denorm problem when the input goes silent after active audio

6 years agoFixes build when DISABLE_FLOAT_API is defined
Pedro Becerra [Thu, 11 Jul 2013 04:00:47 +0000 (00:00 -0400)]
Fixes build when DISABLE_FLOAT_API is defined

Signed-off-by: Jean-Marc Valin <jmvalin@jmvalin.ca>
6 years agoSurround masking tuning
Jean-Marc Valin [Sun, 7 Jul 2013 06:51:07 +0000 (02:51 -0400)]
Surround masking tuning

More conservative surround masking
Disables temporal VBR on surround
Disables tonality on LFE

6 years agoBetter handling of centre for surround masking, bitrate tuning
Jean-Marc Valin [Sun, 7 Jul 2013 06:50:18 +0000 (02:50 -0400)]
Better handling of centre for surround masking, bitrate tuning

Centre channel now contributes sqrt(.5) to both channels
Reduces the stream offset at low bitrate

6 years agoFixes silly fixed-point bug with the surround downmix
Jean-Marc Valin [Sun, 7 Jul 2013 05:11:23 +0000 (01:11 -0400)]
Fixes silly fixed-point bug with the surround downmix

6 years agoPrevents zero-energy on LFE
Jean-Marc Valin [Sun, 7 Jul 2013 04:39:35 +0000 (00:39 -0400)]
Prevents zero-energy on LFE

6 years agoLink libopus itself with libm too
Ron [Tue, 2 Jul 2013 07:31:23 +0000 (17:01 +0930)]
Link libopus itself with libm too

6 years agoMore speech/music detection comments (delayed decision)
Jean-Marc Valin [Tue, 2 Jul 2013 00:19:24 +0000 (20:19 -0400)]
More speech/music detection comments (delayed decision)

6 years agoCommenting the speech/music Markov code
Jean-Marc Valin [Mon, 1 Jul 2013 22:39:34 +0000 (18:39 -0400)]
Commenting the speech/music Markov code

6 years agoDisables temporal VBR for LFE
Jean-Marc Valin [Mon, 1 Jul 2013 20:27:07 +0000 (16:27 -0400)]
Disables temporal VBR for LFE

6 years agoAdds some checks to the surround API
Jean-Marc Valin [Mon, 1 Jul 2013 20:17:01 +0000 (16:17 -0400)]
Adds some checks to the surround API

6 years agoTemporal VBR tuning
Jean-Marc Valin [Mon, 1 Jul 2013 20:12:27 +0000 (16:12 -0400)]
Temporal VBR tuning

Applies temporal VBR up to higher bitrates after IgorC reported it being
useful at 64 kb/s for amount=0.1

6 years agoGet rid of PC_LIBM
Ron [Mon, 1 Jul 2013 17:05:53 +0000 (02:35 +0930)]
Get rid of PC_LIBM

It only existed to not include -lm in the .pc for fixed point builds,
but that is still needed since the float API is still enabled and
will use at least lrint.

6 years agoAdd a couple minor additional decoder tests.
Gregory Maxwell [Sun, 30 Jun 2013 05:20:54 +0000 (22:20 -0700)]
Add a couple minor additional decoder tests.

6 years agoBraces go on the next line.
Gregory Maxwell [Sun, 30 Jun 2013 03:33:32 +0000 (20:33 -0700)]
Braces go on the next line.

6 years agoBrace a number of if statements instead of one-lining them.
Gregory Maxwell [Sun, 30 Jun 2013 03:25:55 +0000 (20:25 -0700)]
Brace a number of if statements instead of one-lining them.

6 years agoFixes some return without va_end in the api, adds tests.
Gregory Maxwell [Sun, 30 Jun 2013 03:06:07 +0000 (20:06 -0700)]
Fixes some return without va_end in the api, adds tests.

Also makes the CTL bad argument handling more consistent to avoid mistakes
 like that in the future.  Also updates the variable duration docs.

6 years agoMake the parallel test mode of automake 1.13 useable
Ron [Sat, 29 Jun 2013 07:06:24 +0000 (16:36 +0930)]
Make the parallel test mode of automake 1.13 useable

6 years agoReplace INCLUDES with AM_CPPFLAGS
Ron [Sat, 29 Jun 2013 06:10:27 +0000 (15:40 +0930)]
Replace INCLUDES with AM_CPPFLAGS

Automake 1.13 complains that INCLUDES is obsoleted now.

6 years agoMore autoconf housekeeping
Ron [Fri, 28 Jun 2013 16:36:53 +0000 (02:06 +0930)]
More autoconf housekeeping

Drop the test for getopt, it's not used anywhere anymore.

Switch the last uses of AC_TRY_COMPILE to AC_COMPILE_IFELSE now.
The former is marked as obsolete, and this will leave no confusion
about which to cut and paste if new tests are added.

Double quote all the parameters to AC_LANG_SOURCE and AC_LANG_PROGRAM.
This is actually required, even if you can get away with not doing it
sometimes, so again set a good example for future changes to follow,
to hopefully avoid people getting bitten harder than they need to be.

Don't bother checking for alloca if we're never going to use it
(ie. if we have C99 variable-size array support).
The test for this is a bit sketchy anyway ...  we separately test for
HAVE_ALLOCA_H and USE_ALLOCA, but the test for USE_ALLOCA depends upon
having alloca.h present, yet the use of these macros in stack_alloc.h
only tests for HAVE_ALLOCA_H inside of a test for USE_ALLOCA.
I'm not going to change this logic right now, since I don't know what
crazy system it was attempting to cater for, though I suspect it was
one that was not using the autoconf build system ... since with the
current test that combination should not be possible to obtain.

Use LT_LIB_M instead of the song and dance with testing for exp().
This should also work for BeOS which is what the exp test was added for.
It also means we don't unconditionally add -lm to everything via LIBS.

Use LIBM now instead of hardcoding -lm everywhere.

Use AS_HELP_STRING to format all option descriptions.

Don't bother to test for doxygen if using it is --disable'd.

Drop the SYMBOL_VISIBILITY export, it isn't used anywhere (we add the
compiler flag to CFLAGS).

6 years agoFixes more MSVC warnings
Jean-Marc Valin [Thu, 27 Jun 2013 18:45:50 +0000 (14:45 -0400)]
Fixes more MSVC warnings

6 years agoFixes MSVC warning
Jean-Marc Valin [Thu, 27 Jun 2013 17:03:33 +0000 (13:03 -0400)]
Fixes MSVC warning

6 years agoAvoid name clash with y0 Bessel function
Jean-Marc Valin [Thu, 27 Jun 2013 08:15:44 +0000 (04:15 -0400)]
Avoid name clash with y0 Bessel function

6 years agoUpdates the Makefile.am EXTRA_DIST for the MSVC reorg
Jean-Marc Valin [Thu, 27 Jun 2013 08:04:48 +0000 (04:04 -0400)]
Updates the Makefile.am EXTRA_DIST for the MSVC reorg

6 years agoFixes MSVC warnings
Jean-Marc Valin [Thu, 27 Jun 2013 07:40:44 +0000 (03:40 -0400)]
Fixes MSVC warnings

6 years agoBig squashed commit of Garf's MSVC reorg:
Jean-Marc Valin [Wed, 26 Jun 2013 18:00:51 +0000 (14:00 -0400)]
Big squashed commit of Garf's MSVC reorg:

commit 77880655f652c54415e4b5f03b5984b951f5267b
Author: Timothy B. Terriberry <tterribe@xiph.org>
Date:   Wed Jun 26 12:23:53 2013 +0200

    Fix version.mk parsing in genversion.bat.

    This was broken by the changes in 0482c82f to make it work with spaces.

    CMD interprets the quotes to mean the argument to FOR is a string,
    not a file name. Instead, we expand the path to a short file name.

Signed-off-by: Gian-Carlo Pascutto <gcp@sjeng.org>
commit 8e97847ecac3a04956747b117fefef14939cbcbe
Merge: 0482c82 3c0aa8f
Author: Gian-Carlo Pascutto <gcp@sjeng.org>
Date:   Wed Jun 26 12:19:41 2013 +0200

    Merge branch 'master' into msvc_reorg

commit 0482c82fea6dd397a7d2ab50dc6cda4965278fc2
Author: Gian-Carlo Pascutto <gcp@sjeng.org>
Date:   Tue May 21 17:59:30 2013 +0200

    Make genversion.bat work in paths with spaces.

commit 3648ec55b7511119735280fc49bc57c7a3b55d6a
Author: Gian-Carlo Pascutto <gcp@sjeng.org>
Date:   Tue May 21 17:42:14 2013 +0200

    Rework output & intermediate dirs for parallel build support.

commit c367a0fcfcbf5551fb65d61365d9b832d7af2b38
Merge: bafb1dd b518b56
Author: Gian-Carlo Pascutto <gcp@sjeng.org>
Date:   Tue May 21 14:17:11 2013 +0200

    Merge branch 'master' into msvc_reorg

    Conflicts:
     win32/VS2010/celt.vcxproj
     win32/VS2010/opus.vcxproj
     win32/VS2010/silk_common.vcxproj
     win32/VS2010/silk_float.vcxproj

commit bafb1dda3fcf6785decdfb42a1f934c7fab0c567
Author: Gian-Carlo Pascutto <gcp@sjeng.org>
Date:   Mon Mar 11 21:25:28 2013 +0100

    Fix two more files to not have the Platform Toolset explicitly set.

commit c69c870201fec7461862fe79731a37866a1810f1
Author: Gian-Carlo Pascutto <gcp@sjeng.org>
Date:   Mon Mar 11 21:08:00 2013 +0100

    Add missing SILK MSVC project files.

commit a0d41fb7f97558c8f812d500b9027c77907c37c0
Author: Gian-Carlo Pascutto <gcp@sjeng.org>
Date:   Fri Mar 8 15:00:51 2013 +0100

    Use the default toolset in the MSVC project files.

commit 34aac5b2637ed38c5620f0d876e08db232849933
Author: Gian-Carlo Pascutto <gcp@sjeng.org>
Date:   Fri Mar 8 14:54:41 2013 +0100

    Move the MSVC project files outside the source tree.

    Bring the project structure more in line with the one in libogg.

6 years agoAdds "temporal VBR", which should be mostly useful at low bitrate.
Jean-Marc Valin [Tue, 25 Jun 2013 18:10:27 +0000 (14:10 -0400)]
Adds "temporal VBR", which should be mostly useful at low bitrate.

The idea is to allow more noise than the default in quiet segments.
Right now, this hurts for high bitrates , so it's gradually reduced
as the rate goes up, until it's completely disabled above 68 kb/s.

6 years agoMakes dual_inner_prod() more generic to increase its use
Jean-Marc Valin [Mon, 17 Jun 2013 20:37:41 +0000 (16:37 -0400)]
Makes dual_inner_prod() more generic to increase its use

6 years agoEnable SSE for MSVC
MuldeR [Mon, 17 Jun 2013 18:09:40 +0000 (14:09 -0400)]
Enable SSE for MSVC

Signed-off-by: Jean-Marc Valin <jmvalin@jmvalin.ca>
6 years agoOnly enable non-multiple of 4 in comb_filter_const() when CUSTOM_MODES is on
Jean-Marc Valin [Mon, 17 Jun 2013 18:02:10 +0000 (14:02 -0400)]
Only enable non-multiple of 4 in comb_filter_const() when CUSTOM_MODES is on

6 years agoUse __declspec(dllexport) on mingw build.
Jacek Caban [Mon, 17 Jun 2013 17:09:57 +0000 (10:09 -0700)]
Use __declspec(dllexport) on mingw build.

Signed-off-by: Gregory Maxwell <greg@xiph.org>
6 years agoSSE optimization of comb_filter()
Jean-Marc Valin [Mon, 17 Jun 2013 07:58:16 +0000 (03:58 -0400)]
SSE optimization of comb_filter()

Should make it easy to adapt to other architectures.

6 years agoAvoids symbol clashes with Speex (pitch_xcorr) and libm (y1)
Jean-Marc Valin [Mon, 17 Jun 2013 04:44:12 +0000 (00:44 -0400)]
Avoids symbol clashes with Speex (pitch_xcorr) and libm (y1)

6 years agoConverts denormalise_bands() to use 16-bit multiplications
Jean-Marc Valin [Mon, 17 Jun 2013 01:56:41 +0000 (21:56 -0400)]
Converts denormalise_bands() to use 16-bit multiplications

6 years agoMoves log2Amp inside denormalise_bands() and get rid of bandE[]
Jean-Marc Valin [Mon, 17 Jun 2013 00:24:52 +0000 (20:24 -0400)]
Moves log2Amp inside denormalise_bands() and get rid of bandE[]

Also get rid of the MSE measurement code which is outdated and no longer useful

6 years agoDon't call denormalise_bands() on silence
Jean-Marc Valin [Sun, 16 Jun 2013 19:40:10 +0000 (15:40 -0400)]
Don't call denormalise_bands() on silence

6 years agoSplit cwrsi() by pulses vs. dimensions.
Timothy B. Terriberry [Sat, 15 Jun 2013 06:57:19 +0000 (23:57 -0700)]
Split cwrsi() by pulses vs. dimensions.

This lets us cut out a bunch of work in the large _n, small _k case
 where most of the dimensions won't have any pulses.
It also gets rid of all remaining usage of CELT_PVQ_U() in cwrsi(),
 leaving just a single test instead of lots of mins and maxes, and
 makes a bunch of the jump threading more obvious.

This is a 1.6% decoder speedup on a 96 kbps comp48-stereo encode on
 a Cortex A8.

6 years agoFurther speedup in cwrsi() by using the special case for n=2
Timothy B. Terriberry [Sat, 15 Jun 2013 06:01:03 +0000 (02:01 -0400)]
Further speedup in cwrsi() by using the special case for n=2

6 years agoFurther optimization to cwrsi()
Timothy B. Terriberry [Sat, 15 Jun 2013 05:35:23 +0000 (01:35 -0400)]
Further optimization to cwrsi()

Makes it possible to skip the first loop in some cases.

6 years agoSplits cwrsi() inner loop in two to avoid the min/max and some load chains
Jean-Marc Valin [Sat, 15 Jun 2013 04:42:38 +0000 (00:42 -0400)]
Splits cwrsi() inner loop in two to avoid the min/max and some load chains

6 years agoSSE optimization of remove_doubling()
Jean-Marc Valin [Fri, 14 Jun 2013 03:51:58 +0000 (23:51 -0400)]
SSE optimization of remove_doubling()

Should be trivial to adapt for Neon.

6 years agoFixes stupid tf calibration bugs introduced/exposed in f77410d
Jean-Marc Valin [Thu, 13 Jun 2013 19:28:53 +0000 (15:28 -0400)]
Fixes stupid tf calibration bugs introduced/exposed in f77410d

6 years agoAdds a quick hack to replace the normal calls with the multistream version.
Jean-Marc Valin [Thu, 13 Jun 2013 19:06:42 +0000 (15:06 -0400)]
Adds a quick hack to replace the normal calls with the multistream version.

6 years agoMoves VBR calculations to a separate function.
Jean-Marc Valin [Mon, 10 Jun 2013 07:30:01 +0000 (03:30 -0400)]
Moves VBR calculations to a separate function.

Does not change the behaviour of the VBR code in most cases. The only
exception is that the VBR offset is now taken into accound in the base_rate,
which will have a (very minor) impact on CVBR at low rate.

6 years agoDrop the stdint size tests that we never use anywhere
Ron [Sat, 8 Jun 2013 15:27:02 +0000 (00:57 +0930)]
Drop the stdint size tests that we never use anywhere

These were probably cribbed from libogg, but we don't use them here,
opus_types.h instead has a list of hardcoded arch definitions.

6 years agoFixes fixed-point on x86 (no SSE).
Jean-Marc Valin [Fri, 7 Jun 2013 11:21:41 +0000 (07:21 -0400)]
Fixes fixed-point on x86 (no SSE).

6 years agoImproved SSE version of xcorr_kernel()
John Ridges [Fri, 7 Jun 2013 03:12:57 +0000 (23:12 -0400)]
Improved SSE version of xcorr_kernel()

The loop no longer reads past its buffer and is slightly faster.
Also fixes RESTORE_STACK in celt_iir().

6 years agoForgot to add assembly file
Jean-Marc Valin [Thu, 6 Jun 2013 16:45:17 +0000 (12:45 -0400)]
Forgot to add assembly file