Commit graph

304 commits

Author SHA1 Message Date
Wim Taymans
5cf367ae57 audioconvert: rework audioconvert
Rewrite audioconvert to try to make it more clear what steps are
executed during conversion.
Add passthrough step that just does a memcpy when possible.
Add ORC optimized dither and quantization functions.
Implement noise-shaping on S32 samples only and allow for arbitrary
noise shaping coefficients if we want this later.
2015-10-30 17:51:47 +01:00
Wim Taymans
e1569ce76a channelmix: fix up API a little
don't use gpointer * for something that should be gpointer.
2015-10-30 17:51:47 +01:00
Wim Taymans
26d469a04b audioquantize: make helper for add with saturation 2015-10-30 17:51:47 +01:00
Wim Taymans
cd6c29e071 audioconvert: make the quantizer a reusable object
Turn the quantizer into a reusable object.
2015-10-28 11:36:18 +01:00
Wim Taymans
8fc2569328 audioconvert: make the channel mixer a separate reusable object
A first attempt at making the channel mixer a separate object.
2015-10-28 11:36:18 +01:00
Wim Taymans
8d4cd51e59 audioquantize: fix 8-pole noise shaping
Fix the 8-pole noise shaping error update. We were mixing errors from
different channels.
2015-10-28 11:36:18 +01:00
Sebastian Dröge
b4afaee8c0 audioconvert: Update disted orc files 2015-10-23 19:13:05 +03:00
Wim Taymans
2b626a5adf audioconvert: use pack/unpack functions
Rework the converter to use the pack/unpack functions
Because the unpack functions can only unpack to 1 format, add a separate
conversion step for doubles when the unpack function produces int.
Do conversion to S32 in the quantize function directly.
Tweak the conversion factor for doing float->int conversion slightly to
get the full range of negative samples, use clamp to make sure we don't
exceed our int range on the positive axis (see also )
2015-10-23 16:58:17 +02:00
Sebastian Dröge
010e35afa7 audioconvert: Also copy metas if their API has no tags attached to it
This is the default basetransform behaviour, being more strict than that
is not really useful.
2015-06-29 13:06:49 +02:00
Mathieu Duponchelle
88484399c5 audioconvert: copy metadata that only has the "audio" tag.
https://bugzilla.gnome.org/show_bug.cgi?id=750406
2015-06-04 19:16:40 +02:00
Tim-Philipp Müller
ec5c93f169 docs: update element example pipelines
- gst-launch -> gst-launch-1.0
- use autoaudiosink and audiovideosink more often
- review pipeline examples and descriptions
2015-05-10 11:38:19 +01:00
Luis de Bethencourt
69f66aff9e Rename property enums from ARG_ to PROP_
Property enum items should be named PROP_ for consistency and readability.
2015-04-27 11:27:00 +01:00
Tim-Philipp Müller
c680e324bc Remove obsolete Android build cruft
This is not needed any longer.
2015-04-26 18:42:34 +01:00
Bernhard Miller
1c01b50ada audioconvert: fix mixed usage of gint and gint32 in int matrix
This is a fixup for b2db18cda2
audioconvert: avoid float calculations when mixing integer-formatted channels

The int matrix was using gint and gint32 synonymously, which can theoretically
cause problems if gint and gint32 are actually different types.

https://bugzilla.gnome.org/show_bug.cgi?id=747005
2015-04-15 12:18:03 +02:00
Bernhard Miller
b2db18cda2 audioconvert: avoid float calculations when mixing integer-formatted channels
The patch calculates a second channel mixing matrix from the current one. The
matrix contains the original values * (2^10) as integers. This matrix is used
when integer-formatted channels are mixed.

On a ARM Cortex-A8, single core, 800MHz this improves performance in a
testcase from 29s to 9s for downmixing 6 channels to stereo.

https://bugzilla.gnome.org/show_bug.cgi?id=747005
2015-04-01 07:31:37 -07:00
Ilya Konstantinov
3dc3aa4e3b audioconvert: Eliminate unsigned quantizers
audio_convert_convert unpacks to default format (signed) before calling
quantize, and the unsigned variants were equivalent to signed anyway,
so we just get rid of them.
2015-03-24 16:52:07 +01:00
Ilya Konstantinov
7b398701cf audioconvert: Avoid int division in quantization
Since range size is always 2^n, we can simply use modulo (implemented
with a bitmask).

The previous implementation used 64-bit integer division, which is
done in software on ARMv7. Although the divisor was constant, the
division could not be transformed into "multiplication by magic number"
since the dividend was 64-bit.

The now-unused and not-so-fast gst_fast_random_(u)int32_range functions
were removed.

Also, implementing bug fixes:

1) ADD_DITHER_TPDF_HF_I no longer discards bias.

2) We change TPDF's noise range to be the same as RPDF's. Previously,
RPDF's noise ranged:
  { bias - dither, bias + dither }
while TPDF's noise ranged:
  { bias/2 - dither/2, bias/2 + dither/2 - 1 } +
  { bias/2 - dither/2, bias/2 + dither/2 - 1 } =
  { bias - dither, bias + dither - 2 }
Now, both range:
  { bias - dither, bias + dither - 1 }

https://bugzilla.gnome.org/show_bug.cgi?id=746661
2015-03-24 16:52:07 +01:00
Luis de Bethencourt
783204824d orc: update orc files 2015-01-27 13:39:14 +00:00
Sebastian Dröge
2bd4ea6e8e Constify some static arrays everywhere 2015-01-21 09:49:47 +01:00
Sebastian Dröge
68f5350c66 Release 1.3.1 2014-05-03 17:50:10 +02:00
Tim-Philipp Müller
bcb8068e27 docs: remove outdated and pointless 'Last reviewed' lines from docs
They are very confusing for people, and more often than not
also just not very accurate. Seeing 'last reviewed: 2005' in
your docs is not very confidence-inspiring. Let's just remove
those comments.
2014-04-26 23:28:57 +01:00
Tim-Philipp Müller
6c6619a669 audioconvert: never do mixing for 1->1 channel conversions
MONO and NONE position are the same, for example, but in
general there isn't much to do here for such a conversion.

Fixes problem in audioconvert, which would end up using
a mixmatrix when converting between different mono format
because it thinks MONO positioning is different from
unpositioned channels, which is not the case in this
special case. The mixmatrix would end up being 0.0 so
audioconvert would convert to silence samples.

https://bugzilla.gnome.org/show_bug.cgi?id=724509
2014-02-18 10:48:07 +00:00
Matej Knopp
12f85c325d audioconvert: Map buffer as READWRITE if the buffer and memory is writable
and only use the input buffer as temporary buffer in that case.

https://bugzilla.gnome.org/show_bug.cgi?id=709408
2013-10-04 14:02:06 +02:00
Tim-Philipp Müller
0715e73725 audioconvert: improve fixate_format function readability even more
Do the flags comparisons only once and re-use the result.
2013-08-23 19:47:57 +01:00
Tim-Philipp Müller
f317061b9c audioconvert: simplify fixate_format function some more
If we have no output format yet, any format will do. The
!out_info condition existed in every path, so just split
it our for clarity. KISS.
2013-08-23 19:43:14 +01:00
Tim-Philipp Müller
7a481c13ae audioconvert: make fixate function more readable
Use some variables to replace accessor macros to make code
a little bit mor readable.
2013-08-23 19:11:17 +01:00
Tim-Philipp Müller
f448977dbd audioconvert: remove unnecessary deep nesting in fixate function
Makes it easier to read and removes two levels of indentation.
2013-08-23 18:53:48 +01:00
Sebastian Dröge
cebae4514a audioconvert: If we have to lose precision, try to lose as less precision as possible
https://bugzilla.gnome.org/show_bug.cgi?id=706624
2013-08-23 18:52:50 +02:00
Sebastian Dröge
ff5d3313d4 Release 1.1.1 2013-06-05 18:31:27 +02:00
Tim-Philipp Müller
f5c0d61be7 Update disted orc backup files
Generated with 0.4.17 now.
2013-04-22 13:58:33 +01:00
Sebastian Dröge
948a4a3632 gst: Add better support for static plugins 2013-04-15 15:52:58 +02:00
Tim-Philipp Müller
5f59b4f7ee Fix FSF address
https://bugzilla.gnome.org/show_bug.cgi?id=687520
2012-11-03 23:05:09 +00:00
Sebastian Dröge
9e6021fe4b audioconvert: Always prefer the input format if possible
Previously we could've chosen another format with the same
depth even if the input format was possible.

Also make sure to chose according to the order in the
caps.
2012-11-01 16:44:05 +01:00
Sebastian Dröge
bc4389806d audioconvert: Also ignore the SIGNED flag when matching an output format 2012-11-01 14:31:29 +01:00
Rasmus Rohde
c286f8ffa2 audioconvert: Prefer output formats with the same depth or at least a higher depth
Enhance current code to prefer an exact match on sample depth if
possible. Also ignore GST_AUDIO_FORMAT_FLAG_UNPACK when checking
equality on the flags.
2012-11-01 14:29:43 +01:00
Sebastian Dröge
3c1041d5eb Revert "gst: Add better support for static plugins"
This reverts commit d2d79e3bc2,
which was accidentially pushed.
2012-10-24 13:26:26 +02:00
Sebastian Dröge
d2d79e3bc2 gst: Add better support for static plugins 2012-10-24 12:10:44 +02:00
Mark Nauwelaerts
a66ff00908 audioconvert: enhance transforming caps
... so as to preserve input format precision,
and preferably not convert at all.
2012-10-19 16:02:44 +02:00
Tim-Philipp Müller
6842698f0d Purge all references to liboil
And remove unused ffmpegcolorspace tests in the process.

https://bugzilla.gnome.org/show_bug.cgi?id=673285
2012-09-29 11:47:52 +01:00
Tim-Philipp Müller
f7c6aa5abd Release 0.11.94 2012-09-14 02:47:54 +01:00
Mark Nauwelaerts
22d7149ba6 audioconvert: plug leak 2012-09-06 14:02:07 +02:00
Mark Nauwelaerts
88e73f8515 audioconvert: prefer channels of base caps when fixating
... which in turn prefers to preserve input channels when converting.
2012-07-25 15:58:19 +02:00
Wim Taymans
5d3b56e9c4 audioconvert: prefix orc functions with audio_convert_orc_ 2012-07-23 17:24:13 +02:00
Tim-Philipp Müller
3c6a3ad629 Use new gst_element_class_set_static_metadata() 2012-04-10 00:45:16 +01:00
Sebastian Dröge
ad42b16375 gst: Update for GST_PLUGIN_DEFINE() API change 2012-04-05 15:11:05 +02:00
Sebastian Dröge
65307dd132 gst: Update versioning 2012-04-04 14:55:15 +02:00
Wim Taymans
b52c035f13 audioconvert: remove useless transform_ip function 2012-04-02 11:21:26 +02:00
Mark Nauwelaerts
aaf84a941e audioconvert: plug caps leak 2012-03-30 16:56:40 +02:00
Wim Taymans
1982d1ce12 Release 0.11.3 2012-03-22 15:51:39 +01:00
Wim Taymans
25137962ad fix for caps API changes 2012-03-11 19:04:41 +01:00