shithub: opus

e1bce7b3 – Felicia Lim <flim@google.com> authored and Timothy B. Terriberry <tterribe@xiph.org> committed on 2016/07/20 09:37

Fix use_dtx for DISABLE_FLOAT_API

6f260111 – Tristan Matthews <tmatth@videolan.org> authored on 2016/07/19 19:12

Fix include warning

c4027c1d – Tristan Matthews <tmatth@videolan.org> authored on 2016/07/19 18:58

Fix include warning

6fccb4b6 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/07/03 21:06

Adds some smoothing to the energy quantization

b3c9e79a – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/06/21 09:37

Tune SILK rate for hybrid with FEC enabled

2b011b64 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/06/21 08:38

Moving FEC decision to the Opus encoder level (from SILK level)

099fc05e – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/06/20 14:54

Taking loss rate into account in compute_equiv_rate()

7780d4a6 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/06/08 11:19

Retuning the tf_analysis() lambda

14845916 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/06/07 22:33

Fixes equiv_rate for CBR

2ca6df03 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/06/07 22:15

Minor TF cleanup (tf_sum was useless), plus comments

f705e9b5 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/06/06 09:43

Use SPREAD_AGGRESSIVE on non-transient hybrid frames

8229f07d – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/06/06 12:54

Don't use too much temporal resolution on hybrid frames at low rate

92d2492b – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/06/06 12:52

Disable patch_transient_decision() on hybrid since it's causing artifacts

cc344b01 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/06/03 12:27

CBR: set gains to their previons values where we're busting the budget

4e75dd07 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/06/03 11:49

Set pulses to zero if we can't meet rate target

6e124863 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/06/03 11:20

Now that we're preventing huge gains, we no longer need to pretend there's

d8781a0f – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/06/03 10:53

CBR: lock the gain on a subframe when the number of pulses stops going down

0a698d65 – Koen Vos <koenvos@users.noreply.github.com> authored and Jean-Marc Valin <jmvalin@jmvalin.ca> committed on 2016/06/03 10:18

merge back in the LTP gain safety limiter

f78e5080 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/06/03 11:26

CBR: Using the same gain increase for fixed-point as for float

508c5710 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/06/02 15:08

Oops, edited the wrong comment in the previous commit

be5a7da1 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/06/02 14:15

Re-tuning compute_equiv_rate()

e775896e – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/06/02 14:11

Getting rid of the SILK penalty for hybrid CBR.

f6c4f646 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/06/02 13:54

Improving CBR for hybrid mode

de1f92c6 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/06/02 11:50

Making the encoder more aggressive at using wider bandwidths

d60816f7 – Koen Vos <koenvos@users.noreply.github.com> authored and Jean-Marc Valin <jmvalin@jmvalin.ca> committed on 2016/06/02 19:03

comment about reduced codelength component in LTP quantizer error metric

3af77bf4 – Koen Vos <koenvos@users.noreply.github.com> authored and Jean-Marc Valin <jmvalin@jmvalin.ca> committed on 2016/06/02 18:35

avoid double if statement for finding best quantized LTP coefs

2799c53c – Koen Vos <koenvos@users.noreply.github.com> authored and Jean-Marc Valin <jmvalin@jmvalin.ca> committed on 2016/06/01 17:49

increase NSQ decision delay from 32 to 40

ae25203f – Koen Vos <koenvos@users.noreply.github.com> authored and Jean-Marc Valin <jmvalin@jmvalin.ca> committed on 2016/06/01 12:29

reduced noise floor in noise shaping analysis by 2 dB

3e054b8e – Koen Vos <koenvos@users.noreply.github.com> authored and Jean-Marc Valin <jmvalin@jmvalin.ca> committed on 2016/06/01 12:10

increase order of noise shaping filter

f06774c6 – Koen Vos <koenvos@users.noreply.github.com> authored and Jean-Marc Valin <jmvalin@jmvalin.ca> committed on 2016/06/01 09:14

more conservative scaling of LTP corrs; remove assert that checks for negative residual energy; discard CB entries leading to such negative energies

07691f15 – Koen Vos <koenvos@users.noreply.github.com> authored and Jean-Marc Valin <jmvalin@jmvalin.ca> committed on 2016/02/21 18:10

Clean up: alignment of comments

90f8c5ef – Koen Vos <koenvos@users.noreply.github.com> authored and Jean-Marc Valin <jmvalin@jmvalin.ca> committed on 2016/02/21 11:21

Clean up: replace tabs by spaces

6e40eb5d – Koen Vos <koenvos@users.noreply.github.com> authored and Jean-Marc Valin <jmvalin@jmvalin.ca> committed on 2016/02/21 06:34

removed prefilter

52cfffe5 – Koen Vos <koenvos@users.noreply.github.com> authored and Jean-Marc Valin <jmvalin@jmvalin.ca> committed on 2016/02/20 16:29

slight clean up

8abd50ea – Koen Vos <koenvos@users.noreply.github.com> authored and Jean-Marc Valin <jmvalin@jmvalin.ca> committed on 2016/02/19 10:38

undo some loop unrolling

6adea4b9 – Koen Vos <koenvos@users.noreply.github.com> authored and Jean-Marc Valin <jmvalin@jmvalin.ca> committed on 2016/02/19 06:11

pitch analysis (in FIX) now scales the input down at the start, instead of at every stage

fc7ce7d5 – Koen Vos <koenvos@users.noreply.github.com> authored and Jean-Marc Valin <jmvalin@jmvalin.ca> committed on 2016/02/19 06:04

minor clean up of pointer offsets

8c9d4186 – Koen Vos <koenvos@users.noreply.github.com> authored and Jean-Marc Valin <jmvalin@jmvalin.ca> committed on 2016/02/18 16:01

simplified computation of LTP coefs

4f5557c3 – Felicia Lim <flim@google.com> authored and Jean-Marc Valin <jmvalin@jmvalin.ca> committed on 2016/03/30 11:46

Attenuate SILK PLC gain only for unvoiced speech

0e69e4ee – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/04/16 23:20

We should now be able to be less conservative when using CBR

894f7294 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/04/19 23:37

Makes the encoder more aggressive about meeting the rate target

89857b5c – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/04/18 09:39

Using more fine-grained complexity settings for SILK

691f6f93 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/05/31 16:49

Fixing bandwidth selection hysteresis

78fc664c – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/05/31 13:54

Boosting the high-band bitrate on transients for hybrid mode

34da05b3 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/04/19 12:21

Fixes signed integer overlof in silk_ADD_POS_SAT32()

33f18a20 – Felicia Lim <flim@google.com> authored and Jean-Marc Valin <jmvalin@jmvalin.ca> committed on 2016/05/16 11:52

Turn on DTX only when the noise is sufficiently quiet

36481346 – Felicia Lim <flim@google.com> authored and Jean-Marc Valin <jmvalin@jmvalin.ca> committed on 2016/05/16 11:29

New DTX that works in all modes (SILK/CELT/HYBRID)

425e8a9b – Felicia Lim <flim@google.com> authored and Jean-Marc Valin <jmvalin@jmvalin.ca> committed on 2016/05/16 11:11

Skip analysis if the frame contains only digital silence

e437424c – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/05/21 20:52

NLSF_VQ_MAX_SURVIVORS is no longer useful

8daa4d37 – Koen Vos <koenvos@users.noreply.github.com> authored and Jean-Marc Valin <jmvalin@jmvalin.ca> committed on 2016/02/21 09:27

take advantage of more efficient NLSF quantization by reducing number of survivors in search

eb2c8f38 – Koen Vos <koenvos@users.noreply.github.com> authored and Jean-Marc Valin <jmvalin@jmvalin.ca> committed on 2016/02/21 08:52

NLSF_del_dec_quant minor clean up

c462ae92 – Koen Vos <koenvos@users.noreply.github.com> authored and Jean-Marc Valin <jmvalin@jmvalin.ca> committed on 2016/02/21 08:47

NLSF_encode now also uses NLSF weights from table

aa7e1748 – Koen Vos <koenvos@users.noreply.github.com> authored and Jean-Marc Valin <jmvalin@jmvalin.ca> committed on 2016/02/21 08:31

NLSF VQ now uses absolute error, and prediction

2955f68b – Koen Vos <koenvos@users.noreply.github.com> authored and Jean-Marc Valin <jmvalin@jmvalin.ca> committed on 2016/02/21 08:16

NLSF decoding now uses tables for NLSF weights

7c645606 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/02/19 05:44

Implement sum_sqr_shift() using two passes with no branch inside the loops

fe4d91c2 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/05/11 15:17

Simplifying fast_atan2f()

61714e9e – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/04/22 20:34

Quality: Increase CELT rate for voiced frames in hybrid mode

45773d83 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/04/27 00:04

Better modeling of the effect of CBR and complexity

ecfdb286 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/04/22 12:13

Re-enabling high-band attenuation for hybrid mode

d68e809b – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/01/07 10:49

Quality: Retrained classifier MLP with better data

7e0ca433 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2015/12/25 08:12

Using "hybrid" flag instead of "start!=0"

64dd0c03 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2015/12/24 09:45

Quality: Making the encoder more aggressive at using wider bandwidths

fb1d3165 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2015/12/25 08:40

Quality: Removes high-band attenuation in hybrid mode -- at least for now

66f3ef31 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2015/12/24 08:03

Quality: Adds SILK rate interpolation function

bcd6abad – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2015/12/06 05:50

Quality: Makes real CELT VBR work for hybrid too

0247d34d – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2015/12/05 09:31

Quality: Forces trim to 5 on hybrid mode

8dcf4acc – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2015/11/27 14:10

Quality: Encoding music in full-band stereo from 32 kb/s

59618a5f – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2015/11/16 18:03

Quality: removes VBR attenuation at low bitrate

c1acbe08 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/07/15 12:03

Adds checksum for 1.1.3

aa32042a – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/07/15 10:29

Slightly increase the safety margin for opus_pcm_soft_clip()

bca70b87 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/07/15 06:57

Fixes opus_pcm_soft_clip() returning values slightly larger than +/-1 with -ffast-math

6be769c3 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/07/15 06:55

Apply 02f3ac225f99 fix for divide-by-zero to the SSE4.1 version of silk_burg_modified()

1a94b2b8 – Jonathan Lennox <jonathan@vidyo.com> authored and Timothy B. Terriberry <tterribe@xiph.org> committed on 2016/07/08 14:30

Fix RTCD on ARM with Neon Intrinsics but not ASM.

37cce280 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/07/08 10:21

bump static PACKAGE_VERSION string in version.mk

03fb50eb – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/07/08 10:20

bump OPUS_LT_REVISION

1c311423 – Mark Harris <mark.hsj@gmail.com> authored on 2016/07/08 07:08

Packets indicating DTX may be up to 2 bytes

5eb270d9 – Ralph Giles <giles@mozilla.com> authored on 2016/07/08 06:23

Test common configure switches on travis.

d02ec93f – Ralph Giles <giles@mozilla.com> authored on 2016/07/08 06:10

Add a minimal travis continuous integration script.

bcea233f – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/07/07 07:54

Makes CBR calculations more accurate for 60-ms frames

eff72e47 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/07/06 22:33

Respect CBR even at ridiculourly low rate

e959f0d5 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/07/06 21:23

Prevents an invalid ToC to be generated at ridiculously low rate

4b1f45e4 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/07/06 18:25

make len==0 return OPUS_INVALID_PACKET rather than OPUS_BAD_ARG for opus_packet_parse_impl()

5e50f1f9 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2016/07/06 17:41

Check for len<1 in opus_packet_parse_impl()

096f5d37 – Timothy B. Terriberry <tterribe@xiph.org> authored on 2016/06/18 06:05

Remove tabs from source code.

05d3e794 – Jonathan Lennox <jonathan@vidyo.com> authored and Timothy B. Terriberry <tterribe@xiph.org> committed on 2015/11/20 18:03

Add Aarch64 intrinsic for SIG2WORD16.

512f17a2 – Jonathan Lennox <jonathan@vidyo.com> authored and Timothy B. Terriberry <tterribe@xiph.org> committed on 2015/11/20 18:03

Add Aarch64 intrinsics for saturated add/subtract.

87973603 – Jonathan Lennox <jonathan@vidyo.com> authored and Timothy B. Terriberry <tterribe@xiph.org> committed on 2015/11/20 18:03

Add configure check for Aarch64-specific Neon intrinsics.

87c670db – Jonathan Lennox <jonathan@vidyo.com> authored and Timothy B. Terriberry <tterribe@xiph.org> committed on 2015/11/20 18:03

Clean up formatting of configure output for ARM intrinsics detection.

81b699c6 – Jonathan Lennox <jonathan@vidyo.com> authored and Timothy B. Terriberry <tterribe@xiph.org> committed on 2015/11/20 18:03

Add OPUS_FAST_INT64 definition of silk_SMULWT.

c59c138d – Jonathan Lennox <jonathan@vidyo.com> authored and Timothy B. Terriberry <tterribe@xiph.org> committed on 2015/11/20 18:03

Explicitly cast results of silk OPUS_FAST_INT64 macros back to opus_int32.

d6a44c55 – Jonathan Lennox <jonathan@vidyo.com> authored and Timothy B. Terriberry <tterribe@xiph.org> committed on 2015/11/20 18:03

Add OPUS_FAST_INT64 flavors of celt/fixed_generic.h macros.

8da4f849 – Jonathan Lennox <jonathan@vidyo.com> authored and Timothy B. Terriberry <tterribe@xiph.org> committed on 2015/11/20 18:03

Move OPUS_FAST_INT64 definition to celt/arch.h.

92f9c82a – Jonathan Lennox <jonathan@vidyo.com> authored and Timothy B. Terriberry <tterribe@xiph.org> committed on 2015/11/20 18:03

Clean up some intrinsics-related wording in configure.

6d279023 – Jonathan Lennox <jonathan@vidyo.com> authored and Timothy B. Terriberry <tterribe@xiph.org> committed on 2015/11/20 18:03

Enable intrinsics by default.

7017f550 – Jonathan Lennox <jonathan@vidyo.com> authored and Timothy B. Terriberry <tterribe@xiph.org> committed on 2015/12/22 14:21

Add Neon fixed-point implementation of xcorr_kernel.

7f1c6fa6 – Jonathan Lennox <jonathan@vidyo.com> authored and Timothy B. Terriberry <tterribe@xiph.org> committed on 2015/12/22 14:21

Add Neon intrinsics for Silk noise shape feedback loop.

81a19427 – Jonathan Lennox <jonathan@vidyo.com> authored and Timothy B. Terriberry <tterribe@xiph.org> committed on 2015/12/22 14:21

Apply Neon short prediction optimization to silk_noise_shape_quantizer_del_dec.

44e8e738 – Jonathan Lennox <jonathan@vidyo.com> authored and Timothy B. Terriberry <tterribe@xiph.org> committed on 2015/12/22 14:21

Add Neon intrinsics for Silk noise shape quantization.

fdb3b3ac – Jonathan Lennox <jonathan@vidyo.com> authored and Timothy B. Terriberry <tterribe@xiph.org> committed on 2015/12/22 14:21

Add named constants for ARM architecture variants.

920ff718 – Jonathan Lennox <jonathan@vidyo.com> authored and Timothy B. Terriberry <tterribe@xiph.org> committed on 2015/11/20 18:02

Enable Neon intrinsics for aarch64.