shithub: opus

Info  •  Files  •  Log  •  Branches

ce06772d – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/05/23 13:01

Better MDCT/combfilter accuracy with MULT32_32_P31

c8eb0623 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/06/11 07:02

Fix include path

4d269fc9 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/06/06 12:08

Saturate dc_reject() input to prevent wrap-around

38e08cf7 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/06/06 10:59

Prevent wrap-around in silk_resampler_down2_hp()

d17b5dfd – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/06/05 13:54

Silence FFT wrap-around

74cdcaec – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/06/05 07:22

Fix potential overflow in IMDCT headroom

462c50d3 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/06/04 10:00

Disabling the comb filter when it's not useful

9840debe – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/05/30 06:08

Improve transient_analysis() fixed-point accuracy

d7eceaea – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/05/29 11:29

Simplify tone_detect()

e082ddc1 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/05/29 08:07

Moving transient analysis before the prefilter

2941f08a – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/05/13 08:29

Adaptive IMDCT scaling to maximize accuracy

4df773d0 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/05/06 12:41

Improve transient_analysis() fixed-point accuracy

08bcc6e4 – Mikhael Skvortsov <mikhael.skvortsov@gmail.com> authored and Jean-Marc Valin <jmvalin@jmvalin.ca> committed on 2025/04/26 11:24

Optimizations for Xtensa LX7

a3890352 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/04/25 11:58

Set opus_demo LSB depth according to input format

f37dcf55 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/04/24 16:43

Fix msan uninitialized psEncCtrl->LTP_scale_Q14

6a45b767 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2025/04/21 07:24

Add skewed split for fine-tuning decoder

cb7cf92a – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2025/04/21 07:23

DRED: Add lambda schedule for first epochs

a41a344a – Sandor Vegh <sandorzsombor.vegh@arm.com> authored and Jean-Marc Valin <jeanmarcv@google.com> committed on 2024/10/03 12:08

Arm: Speed up -1..1 soft clipping with Neon

62ea8ea1 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/04/16 06:58

Don't limit multi-frame packets to 1276 bytes

d8d77633 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/04/04 10:46

Fix decoder gain for res24

f0bf32a5 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/04/04 10:44

Fix projection encoder integer overflow

60b8f5b8 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/04/03 10:45

Add SHL32_ovflw() and use it in IMDCT

a013aa04 – Jan Buethe <jan.buethe@gmx.net> authored on 2025/04/06 13:29

deprecating _DELAY from conv1d layer dump

df02d25f – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2025/03/29 07:12

Updated FARGAN model with softquant

af6dbd84 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2025/03/28 15:28

Adds --chunks-per-offset option to train_rdovae.py

1ca6933a – Jan Buethe <jan.buethe@gmx.net> authored and Jean-Marc Valin <jmvalin@jmvalin.ca> committed on 2025/03/08 08:03

added soft quantization to RDOVAE and FARGAN

ebccedd9 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2025/03/28 15:25

Fix rdovae exporting

bb564fab – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/03/28 15:19

UBSan fixes for opus_demo

1c370855 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/03/26 11:04

Forgot one wraparound to ignore

e6dd7a3d – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/03/26 10:59

Do not run valgrind on -ffast-math builds

d4494e6e – Sandor Zsombor Vegh <sandorzsombor.vegh@arm.com> authored and Jean-Marc Valin <jeanmarcv@google.com> committed on 2024/09/11 10:00

Arm: Speed up FLOAT2INT16 conversion with Neon

edffe56b – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/03/21 12:22

Add PSHR32_ovflw() and use it in IMDCT

fba923d8 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/03/21 06:41

Fix fixed-point gain wrap-around on invalid packet

b5aad6a2 – Jean-Marc Valin <jmvalin@jmvalin.ca> authored on 2025/03/18 10:01

Better model building instructions

55513e81 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/03/18 09:41

Fix multiple unsequenced modifications

4a344aff – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/03/17 10:15

Work around a clang VLA issue

5a29ca93 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/03/14 12:06

Looser error bound for celt_log2()

738c29fc – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/03/14 10:04

Fix numerical issue in tone detector

2169f64c – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/02/27 05:13

Increase accuracy of QEXT twiddle factors/window

e1d41b6e – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/02/27 15:40

More warnings about unsigned integer overflows

8b4f3f64 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/02/27 15:13

Fix warnings about unsigned integer overflows

377cfaa2 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/02/28 11:58

Delay forward MDCT scaling to improve accuracy

5beef650 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/02/26 06:05

Gradual downshift in FFT to improve accuracy

03e0d3a6 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/02/27 10:50

Fix MDCT headroom calculation from 38e535a1

030074f9 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/02/28 09:59

Fix overflow in fixed-point constant

c79a9bd1 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/05/28 09:03

Support 32-bit window and comb filter

38e535a1 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/07/04 21:28

Improve MDCT scaling to maximize accuracy

3116093d – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/05/27 12:29

Support for full 32-bit FFT/MDCT twiddles/scaling

ea3eb56a – Yunho Huh <yunho@google.com> authored and Jean-Marc Valin <jeanmarcv@google.com> committed on 2024/12/10 16:54

Update the celt_exp2_db to match the new floating point celt_exp2.

255f0130 – Yunho Huh <yunho@google.com> authored and Jean-Marc Valin <jeanmarcv@google.com> committed on 2024/11/02 09:02

Improve exp2's precision to 20-24 bits.

7fa23b4e – Yunho Huh <yunho@google.com> authored and Jean-Marc Valin <jeanmarcv@google.com> committed on 2024/12/09 03:25

Update celt_log2_db fixed point to match new celt_log2 floating point.

e75503be – Yunho Huh <yunho@google.com> authored and Jean-Marc Valin <jeanmarcv@google.com> committed on 2024/11/02 08:27

Improve log2's precision to 20-24 bits.

4ba06d9e – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/06/28 08:54

Increasing log gain resolution to DB_SHIFT=24

abd512f7 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/06/26 09:37

Defining MAXG/MING/GCONST for log gain

be37d866 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/07/18 07:15

Cleanup: use celt_glog for log of the gain

100228e3 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/06/25 21:35

Cleaning up the use of DB_SHIFT

37388eb5 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/06/29 23:16

Using 32 bits for fixed-point band split gains

45a4e4e8 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/01/31 05:32

Add more formats to opus_demo

3399621b – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/06/13 06:55

Adds 24-bit API for projection encoder/decoder

d04f59b4 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/06/12 12:25

Cleaning up projection encode/decoder functions

d4744e0b – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/06/12 12:15

Adds 24-bit multistream encode/decode API

f7495d4a – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/06/12 11:24

Cleanup multistream encode/decode API functions

3171ed6f – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/06/12 11:21

Cleanup Opus encode/decode API functions

ec06c2ae – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/06/12 10:43

Cleanup Opus custom API

977ee016 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/06/12 08:57

Adds 24-bit Opus encoder/decoder API

8b16ce90 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/06/12 07:48

Adds 24-bit API to Opus custom

58208f44 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/06/11 22:06

Refactoring analysis downmix to use sig scaling

82485dd2 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/06/10 20:55

Preserving 24-bit accuracy for fixed-point encoder

a4854afa – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/06/08 17:46

Preserving 24-bit accuracy for fixed-point decoder

4141c4d5 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/06/28 08:56

Extra bounds check in case

0164bdb3 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/06/29 21:49

More accurate fixed-point normalise_bands()

cdba712c – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/06/29 21:47

More accurate fixed-point compute_band_energies()

e9d7842c – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/06/29 21:42

More accurate fixed-point sqrt() approximation

6f4f3e89 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/06/28 15:54

Disable TF analysis on tones

86101c0e – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/06/21 12:52

Improve encoder allocation on tones

3b68a486 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/06/20 21:46

Add LPC-based tone detection

148ccd81 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/02/12 06:19

Abort if we overflow the pseudostack

360fd071 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/01/22 10:29

Test corrupted bitstreams in test_opus_custom

6f9cb7aa – Jean-Marc Valin <jeanmarcv@google.com> authored on 2025/01/24 05:38

C90 fixes: Removing declarations after statement

734aed05 – Timothy B. Terriberry <tterribe@xiph.org> authored and Jean-Marc Valin <jeanmarcv@google.com> committed on 2025/02/10 05:54

Remove commented-out code.

60d6c560 – Timothy B. Terriberry <tterribe@xiph.org> authored and Jean-Marc Valin <jeanmarcv@google.com> committed on 2024/11/03 21:29

Always use L=0 for last repeated long extension.

b49c07d1 – Timothy B. Terriberry <tterribe@xiph.org> authored and Jean-Marc Valin <jeanmarcv@google.com> committed on 2024/07/26 02:19

Advance frame number after an L=0 RTE.

60b527ae – Timothy B. Terriberry <tterribe@xiph.org> authored and Jean-Marc Valin <jeanmarcv@google.com> committed on 2024/07/28 13:02

Check regenerating then parsing random extensions.

10ceaedb – Timothy B. Terriberry <tterribe@xiph.org> authored and Jean-Marc Valin <jeanmarcv@google.com> committed on 2024/07/25 16:48

Encoder implementation of Repeat These Extensions.

58bf8b41 – Timothy B. Terriberry <tterribe@xiph.org> authored and Jean-Marc Valin <jeanmarcv@google.com> committed on 2024/07/24 10:28

Decoder implementation of Repeat These Extensions.

5c13840b – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/07/04 17:25

Saturate de-emphasis so we can increase SIG_SAT

7b317ceb – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/07/04 17:32

celt_sqrt() can return 32-bit value 32768

a6d29ed5 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/07/05 09:20

Fix undefined left shift of negative value

140e0982 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/07/31 07:19

Fix run-time warning in NSQ_del_dec_avx2()

1b208622 – Jeff Peil <jeffpeil@google.com> authored and Jean-Marc Valin <jeanmarcv@google.com> committed on 2024/09/20 11:16

Add OpusCustom Test Script

5ee25fe2 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/09/27 09:15

Fix low-bitrate encode/decode spreading mismatch

95784cbc – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/08/02 07:54

Fix OpusCustom encoder busting at low bitrate

5e5e7f69 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/08/01 11:46

Fix bitrate management for OpusCustom

fcd513e7 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/09/16 10:50

Fix anti-collapse RESYNTH mismatch for mono

8bead748 – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/09/24 06:17

Don't crash OpusCustom when neural PLC is enabled

8e6e049f – Jean-Marc Valin <jeanmarcv@google.com> authored on 2024/10/28 08:33

Refactoring opus_custom_demo

7db26934 – Jan Buethe <jan.buethe@gmx.net> authored on 2024/10/16 06:51

fix for buffer size calculation in osce.c

ff6dea5e – Ralph Giles <giles@thaumas.net> authored on 2024/09/11 07:44

fix spelling error in docs

935015f3 – IOhannes m zmölnig <umlaeute@debian.org> authored and Ralph Giles <giles@thaumas.net> committed on 2022/06/18 07:42

fix spelling errors in docs

6dabc64a – Timothy B. Terriberry <tterribe@xiph.org> authored on 2024/07/25 07:39

Add extension iterator.