RSS feed
54aab422
– Jean-Marc Valin <
[email protected]>
authored
on 2008/12/06 02:52
Cleaned up some junk left from the last allocator change
c7635b45
– Jean-Marc Valin <
[email protected]>
authored
on 2008/12/04 18:26
New VQ search works in fixed-point even though there's still some float ops left
a733f08e
– Jean-Marc Valin <
[email protected]>
authored
on 2008/12/04 17:52
Experimental code to improve both the speed and accuracy of the VQ search
6cde5dd2
– Jean-Marc Valin <
[email protected]>
authored
on 2008/12/04 16:21
alg_quant() now handles the sign of X[] separately from the quantisation
920648e0
– Jean-Marc Valin <
[email protected]>
authored
on 2008/11/28 09:43
Getting rid of more stuff left from the Speex draft
de1fd287
– Jean-Marc Valin <
[email protected]>
authored
on 2008/11/27 18:34
Fixes a denormalisation issue occuring when multiple packets are lost and the
c6eb40f8
– Gregory Maxwell <
[email protected]>
authored
and Jean-Marc Valin <
[email protected]>
committed
on 2008/11/26 15:16
Celtclient improvements: A usable usage message and correct usage of the jitter buffer.
bc799917
– Jean-Marc Valin <
[email protected]>
authored
on 2008/11/07 17:53
The fold bit can now be used by the encoder and decoder (encoder still needs
a2012d8d
– Jean-Marc Valin <
[email protected]>
authored
on 2008/11/05 16:54
experimental code for choosing whether to apply folding for the high bands
3b34e188
– Gregory Maxwell <
[email protected]>
authored
and Jean-Marc Valin <
[email protected]>
committed
on 2008/10/28 16:28
In some conditions the allocation could run out of bits and attempt to allocate a negative number of pulses.
c5f2a9d9
– Jean-Marc Valin <
[email protected]>
authored
on 2008/10/26 18:00
Latest psychoacoustics work -- still highly experimental
761811d9
– Jean-Marc Valin <
[email protected]>
authored
on 2008/10/18 05:11
re-enable support for resizable buffers in the range coder
8679a80e
– Jean-Marc Valin <
[email protected]>
authored
on 2008/10/18 03:44
Encoder now writes data directly in the user buffer
56cb5cf3
– Jean-Marc Valin <
[email protected]>
authored
on 2008/10/17 17:27
Multiply-free version of the range coder. Haven't yet decided which version to
b0bef856
– Jean-Marc Valin <
[email protected]>
authored
on 2008/10/02 19:52
compute_alloc_cache() isn't defined when STATIC_MODES is.
82595311
– Gregory Maxwell <
[email protected]>
authored
and Jean-Marc Valin <
[email protected]>
committed
on 2008/09/30 14:20
API: Change celt_encode and celt_encode_float so that they take an optional synthesis parameter after the PCM input. If optional_synthesis is null the encoder will be able to save some computation. If optional_synthesis is non-null if will be used to write the encoder's expectation of the decoder's output. Synthesis may alias the input pcm, so calling the encoder with the same buffer twice will achieve the old behavior. Remove 'restrict' from the CTL prototype.
cbaf67e8
– Gregory Maxwell <
[email protected]>
authored
and Jean-Marc Valin <
[email protected]>
committed
on 2008/09/28 00:19
Fix support for sample rates between 32kHz and 40kHz (lower rates are currently rejected, but work okay).
23e654ff
– Gregory Maxwell <
[email protected]>
authored
and Jean-Marc Valin <
[email protected]>
committed
on 2008/09/27 12:20
Allow all even frame sizes, change the selection of number of short MDCTs per frame to be more intelligent.
0ec7c14e
– Jean-Marc Valin <
[email protected]>
authored
on 2008/09/22 06:25
Unb0rked a fixed-point regression caused in 7bb339d
7bb339d9
– Jean-Marc Valin <
[email protected]>
authored
on 2008/09/21 17:11
Split the last pulse out of the alg_quant() main loop.
d125c264
– Jean-Marc Valin <
[email protected]>
authored
on 2008/09/19 13:22
Removed a bunch of divides from the fine energy quantisation
abf5c8ed
– Jean-Marc Valin <
[email protected]>
authored
on 2008/09/19 04:02
Merge branch 'cwrs_speedup' (derf's cwrs changes)
b781877e
– Jean-Marc Valin <
[email protected]>
authored
on 2008/09/12 17:27
Converted everything to 32-bit CWRS (using split after that)
7b0cb4ba
– Jean-Marc Valin <
[email protected]>
authored
on 2008/09/10 20:20
One-at-a-time allocator now uses direct feedback from ec_*_tell().
825ead82
– Jean-Marc Valin <
[email protected]>
authored
on 2008/09/08 20:15
Simplified allocation by doing it one band at a time, with memory
d9b9565a
– Jean-Marc Valin <
[email protected]>
authored
on 2008/08/31 19:34
Added Erik de Castro Lopo's float_cast.h to use lrintf() when available. Also
58b3727f
– Jean-Marc Valin <
[email protected]>
authored
on 2008/08/30 10:53
exp() and log() are faster than pow() and log10() for dB conversion
c994394f
– Jean-Marc Valin <
[email protected]>
authored
on 2008/08/29 20:55
Added signalling bits for enabling/disabling pitch, short blocks, and folding.
5c3bc679
– Jean-Marc Valin <
[email protected]>
authored
on 2008/08/28 19:34
Fixed a bunch of fixed-point overflows on insanely hot signals by changing
8dff923a
– Jean-Marc Valin <
[email protected]>
authored
on 2008/08/13 18:07
Added hooks for tuning pulses and fine energy quant
5eef264f
– Jean-Marc Valin <
[email protected]>
authored
on 2008/08/06 19:06
Solved an encoder/decoder mismatch happening when there's no pulse in the first
4117ff02
– Jean-Marc Valin <
[email protected]>
authored
on 2008/08/05 18:03
Preventing encoder/decoder mismatch when coarse energy budget is busted
504f8c25
– Jean-Marc Valin <
[email protected]>
authored
on 2008/08/05 16:39
some tuning to fine energy allocation and prevent it from going above 7 bits
9c70906f
– Jean-Marc Valin <
[email protected]>
authored
on 2008/08/03 18:07
Changing the allocation algorithm to better take into account the fixed cost
36e6e01a
– Jean-Marc Valin <
[email protected]>
authored
on 2008/08/02 18:25
Merged compute_allocation_table() and compute_energy_allocation_table()
9eba823c
– Jean-Marc Valin <
[email protected]>
authored
on 2008/08/02 16:51
Turns out that the worse case of the vector split is the same as that of the
eb7ce031
– Jean-Marc Valin <
[email protected]>
authored
on 2008/08/02 16:14
fixed a bunch of bugs in the unified allocation code.
c890b58b
– Jean-Marc Valin <
[email protected]>
authored
on 2008/08/01 18:26
Completed the separation of coarse and fine energy quantisation
5c0d486f
– Jean-Marc Valin <
[email protected]>
authored
on 2008/07/24 04:49
Better bit allocation for all frame sizes that are not equal to 256.
1d27ef8b
– Jean-Marc Valin <
[email protected]>
authored
on 2008/07/20 19:30
Re-implemented renormalise_bands() based on renormalise_vector().
6361ad8d
– Jean-Marc Valin <
[email protected]>
authored
on 2008/07/20 19:14
Simplified the folding normalisation by creating renormalise_vector().
2c733067
– Jean-Marc Valin <
[email protected]>
authored
on 2008/07/17 12:22
isolated the folding part of the code that's common in vq.c
7028d62f
– Jean-Marc Valin <
[email protected]>
authored
on 2008/07/10 19:06
Do not encode the short-window bit when not applicable.
bd7de0a7
– Jean-Marc Valin <
[email protected]>
authored
on 2008/07/06 06:57
cleaned up transient_analysis() and replaced the algorithm with a simpler one
f4578196
– Jean-Marc Valin <
[email protected]>
authored
on 2008/07/04 12:47
fixed-point: conversion of pre-echo avoidance now complete.
0ceccb2d
– Jean-Marc Valin <
[email protected]>
authored
on 2008/07/03 16:41
fixed-point: more work on the time window (almost done)
528f4b86
– Jean-Marc Valin <
[email protected]>
authored
on 2008/06/28 23:46
More fixed-point work on the time window -- including conversion of the gain
9375aa4f
– Jean-Marc Valin <
[email protected]>
authored
on 2008/06/27 03:57
Cleaned up the pre-echo avoidance code so it works when compiled as fixed-point
16ca18b9
– Jean-Marc Valin <
[email protected]>
authored
on 2008/06/18 19:44
Automatically choosing the overlap based on the frame size.
12b22484
– Jean-Marc Valin <
[email protected]>
authored
on 2008/06/16 10:13
Implemented two pre-echo avoidance techniques: time-domain pre-emphasis and
b0c153b1
– Jean-Marc Valin <
[email protected]>
authored
on 2008/06/11 14:05
Doing the cwrs split in dimensions should save a few bits.
f892d5f6
– Jean-Marc Valin <
[email protected]>
authored
on 2008/06/10 19:46
Turns out maxK and maxM were also inverted in fits_in64()
679083f4
– Jean-Marc Valin <
[email protected]>
authored
on 2008/06/10 13:23
Implemented split-cwrs for very large codebooks (>64 bits), but still getting
Next