RSS feed
b76888dc
– Jean-Marc Valin <
[email protected]>
authored
on 2014/01/07 16:00
Preventing unnecessary stack use when using a large decode buffer
81af2595
– Jean-Marc Valin <
[email protected]>
authored
on 2014/01/07 11:26
Another aliasing hack in the SILK PLC with SMALL_FOOTPRINT
e17ca256
– Jean-Marc Valin <
[email protected]>
authored
on 2014/01/07 10:27
Don't allocate pulses on the stack when calling the SILK PLC.
b63e7110
– Jean-Marc Valin <
[email protected]>
authored
on 2014/01/07 10:02
Moves CELT PLC pitch search to a separate function to reduce peak stack
c2583620
– Jean-Marc Valin <
[email protected]>
authored
on 2014/01/07 08:46
Delaying allocation of the SILK temporary output buffer to reduce peak stack
9d1b6fef
– Jean-Marc Valin <
[email protected]>
authored
on 2014/01/06 23:32
Moves deemphasis() call out of celt_decode_lost() to reduce peak stack
ad8371d1
– Jean-Marc Valin <
[email protected]>
authored
on 2014/01/06 12:45
Cleaning up leftovers of "freq" in celt_decode_with_ec()
4d07b135
– Jean-Marc Valin <
[email protected]>
authored
on 2014/01/06 12:43
Reduces the decoder stack use by removing the pcm_silk buffer in fixed-point
14ca4ed6
– Jean-Marc Valin <
[email protected]>
authored
on 2014/01/06 04:31
Moves the remains of compute_inv_mdcts() to celt_synthesis()
32454dca
– Jean-Marc Valin <
[email protected]>
authored
on 2014/01/06 04:11
Hack that makes the SMALL_FOOTPRINT CELT decoder use only 4.25 kB of stack.
bdc7b933
– Jean-Marc Valin <
[email protected]>
authored
on 2014/01/06 03:58
Reduces decoder stack usage by only storing one channel of denormalized MDCT
ed01a596
– Jean-Marc Valin <
[email protected]>
authored
on 2014/01/04 16:06
Making exp_rotation1() use MAC16_16(), which saves a few cycles on ARM
ef0eac49
– Jean-Marc Valin <
[email protected]>
authored
on 2014/01/03 18:55
Moving the radix-2 to expose trivial twiddle factors
c8f62a4a
– Jean-Marc Valin <
[email protected]>
authored
on 2013/12/30 19:00
Improving the accuracy of the fixed-point radix-3 and radix-5
05291fd6
– Jean-Marc Valin <
[email protected]>
authored
on 2013/12/29 13:31
Fixed-point: slight accuracy improvement in the comb filter
30f52cbe
– Jean-Marc Valin <
[email protected]>
authored
on 2013/12/29 11:21
Remove a SAVE_STACK that was pasted accidentally in the previous commit
cc344fb8
– Jean-Marc Valin <
[email protected]>
authored
on 2013/12/28 14:10
Slightly improving the accuracy of the fixed-point MDCT downscale
e0c00e27
– Jean-Marc Valin <
[email protected]>
authored
on 2013/12/26 22:16
Commit 99968ab was causing us to allocate too much stack in the MDCT
e43a0abe
– Jean-Marc Valin <
[email protected]>
authored
on 2013/12/26 19:10
Removes the separate 1/8N rotation in the (I)MDCT and unmerges the MDCT sizes
a5e3c8a6
– Jean-Marc Valin <
[email protected]>
authored
on 2013/12/22 21:26
Inverse MDCT no longer requires any scratch space
e2bcb3fe
– Jean-Marc Valin <
[email protected]>
authored
on 2013/12/21 21:17
Reverse the ordering of the FFT stage to optimize a degenerate radix-4 case.
bc13bbaa
– Jean-Marc Valin <
[email protected]>
authored
on 2013/12/20 21:33
Applying the forward FFT gain up-front for fixed-point too
8e02afcf
– Jean-Marc Valin <
[email protected]>
authored
on 2013/12/17 08:56
Fixes a warning about "conversion from '__int64' to 'int'" on MSVC
306d7f5a
– Jean-Marc Valin <
[email protected]>
authored
on 2013/12/15 20:08
fixed-point: slight (but free) accuracy improvement in compute_band_energies()
e0f26246
– Jean-Marc Valin <
[email protected]>
authored
on 2013/12/14 06:07
fixed-point: adds rounding to some shifts to eliminate bias
5752d659
– Jean-Marc Valin <
[email protected]>
authored
on 2013/12/10 19:21
Minor fixed-point accuracy improvements that were completely free
5607d5d1
– Jean-Marc Valin <
[email protected]>
authored
on 2013/12/10 13:10
Annotating pointer arguments with OPUS_RESTRICT and const
32ada84b
– Jean-Marc Valin <
[email protected]>
authored
on 2013/12/10 11:41
Fixes code that was doing arithmetic on a void pointer.
fa85e3bf
– Timothy B. Terriberry <
[email protected]>
authored
on 2013/12/10 03:16
Add a reminder to bump the version.mk version.
d5553e8a
– Jean-Marc Valin <
[email protected]>
authored
on 2013/12/09 21:32
Using OPUS_COPY()/OPUS_CLEAR() in the decoder too
4fda6b01
– Jean-Marc Valin <
[email protected]>
authored
on 2013/12/09 13:05
Using celt_inner_prod() in compute_band_energies()
3039c9bd
– Jean-Marc Valin <
[email protected]>
authored
on 2013/12/09 11:31
Makes silk_inner_prod_aligned() use celt_inner_prod() in fixed-point.
57cd849c
– Jean-Marc Valin <
[email protected]>
authored
on 2013/12/08 21:33
Defining celt_inner_prod() and using it instead of explicit loops.
ff072009
– Jean-Marc Valin <
[email protected]>
authored
on 2013/12/08 18:31
Replaces inline copies and initialization with OPUS_*() macros.
0f869cba
– Jean-Marc Valin <
[email protected]>
authored
on 2013/12/07 22:54
Changes ABS16() and ABS32() to use fabs() in the float build
c94e4bb1
– Jean-Marc Valin <
[email protected]>
authored
on 2013/12/07 22:31
Optimizes encoder NaN detection and clipping by only running them when needed
5626908e
– Jean-Marc Valin <
[email protected]>
authored
on 2013/12/05 11:40
Fixed-point fast-path for normal 48 kHz encoding in celt_preemphasis()
aed1009d
– Jean-Marc Valin <
[email protected]>
authored
on 2013/12/05 08:36
Turns a 16x32 multiply into a 16x16 one in celt_preemphasis().
17dcb270
– Jean-Marc Valin <
[email protected]>
authored
on 2013/12/03 08:31
Revert "Disables assembly optimizations by default because of several problems reported"
2446445b
– Jean-Marc Valin <
[email protected]>
authored
on 2013/12/02 17:17
Disables assembly optimizations by default because of several problems reported
086ea7c7
– Jean-Marc Valin <
[email protected]>
authored
on 2013/11/29 11:47
Whoa! This fixes a pretty bad fixed-point stereo regression
35a44c61
– Jean-Marc Valin <
[email protected]>
authored
on 2013/11/29 07:21
Fixes OPUS_SET_GAIN (was a bug in MULT16_32_P16())
5c02c5ff
– Timothy B. Terriberry <
[email protected]>
authored
on 2013/11/26 16:51
Make celt_pitch_xcorr_edsp() work on ARMv5TE.
2924af4d
– Jean-Marc Valin <
[email protected]>
authored
on 2013/11/25 16:24
Merges surround and stereo intensity stereo thresholds again.
93e10960
– Jean-Marc Valin <
[email protected]>
authored
on 2013/11/24 20:52
Makes SILK surround masking a bit more aggressive in both directions
a47d6f30
– Jean-Marc Valin <
[email protected]>
authored
on 2013/11/24 18:59
Intensity stereo update, partially based on Monty's listening
f4d2c034
– Gregory Maxwell <
[email protected]>
authored
on 2013/11/24 01:49
Put custom_encoder_init decl behind CUSTOM_MODES. More softclip tests.
1bf1f51c
– Jean-Marc Valin <
[email protected]>
authored
on 2013/11/23 21:01
Using a more reasonably SILK surround calibration value
d94ed205
– Jean-Marc Valin <
[email protected]>
authored
on 2013/11/23 20:59
Reducing the thresholds for enabling stereo and higher audio bandwidths
ccf7d9b1
– Jean-Marc Valin <
[email protected]>
authored
on 2013/11/23 20:57
Surround low-passing thresholds set to much lower bitrates
abf91fe9
– Gregory Maxwell <
[email protected]>
authored
on 2013/11/22 06:32
Minor opus_pcm_soft_clip API hardening and tests.
21b131fb
– Jean-Marc Valin <
[email protected]>
authored
on 2013/11/21 09:42
Applies padding to all CBR packets, including PLC and CELT silence
41e8906c
– Jean-Marc Valin <
[email protected]>
authored
on 2013/11/20 14:34
opus_custom_encoder_init() now only needed for custom modes.
854e9bd4
– Jean-Marc Valin <
[email protected]>
authored
on 2013/11/20 08:39
Using the maximum frequency response of pitch taps instead of maxabs
ac31a0af
– Gregory Maxwell <
[email protected]>
authored
on 2013/11/19 03:52
Additional opus_multistream_packet_(un)pad tests.
Next