RSS feed
83006eec
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/21 06:40
fixed-point: using TI intrinsic for celt_ilog2() if available.
b311554c
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/16 19:33
fixed-point: more TI macros. Comments on the existing ones.
b674a278
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/16 17:39
Removed useless functions that caused warnings in skeleton.c
c7e0b76c
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/16 03:55
Making sure not to use the C library calls directly
fb7359a6
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/15 04:04
fixed-point: Fixed a "shift too large" warning by using a 16-bit mul instead
15b9be3a
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/14 13:58
fixed-point: playing it safe. SHL32() now automatically casts input to 32-bit
0773fec4
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/14 12:28
fixed-point: fixed a 16-bit issue in normalise_bands() where Q15ONE was directly
2b35de91
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/13 13:55
fixed-point: Using reciproqual instead of DIV32_16 for find_spectral_pitch()
679f5cc6
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/13 13:39
Enabling warnings and restricting symbol visibility
9d5b4a6f
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/13 07:36
fixed-point: simplification of the gain in mix_pitch_and_residual()
19ae9fc9
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/13 07:18
fixed-point: simplifying the arithmetic in alg_quant()
3ec78b17
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/12 19:00
More accurate sqrt approximation using MULT16_16_Q15() instead of Q14.
887ad9ff
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/12 17:34
Increased accuracy of the reciprocal approximation and made it use
0ac437b8
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/12 14:04
Testcases should now work even when symbols aren't visible in the dso
31b79d19
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/12 13:17
Changed definition of VARDECL to make it more flexible. No actual code change.
d857ac48
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/12 09:26
Using reciprocal approximation instead of full 32-bit division in alg_quant()
980ad384
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/12 07:47
configure update (MIXED_PRECISION and STATIC_MODES can now be used)
3c7ae6c3
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/12 06:42
__GNUC_PREREQ should be causing problems anymore.
72e8003f
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/12 06:16
Added macro definitions for the TI C5x family (untested)
493b38d4
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/11 11:24
Better heuristic for deciding when to use the 32-bit version of the cwrs code.
949902fd
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/11 06:43
const correctness. celt_mode_create() now makes a copy of the static mode
db0141e7
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/10 12:37
Changed some of the mode data from int to celt_int16_t to save memory.
6a0d97eb
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/10 12:10
No longer include redundent entries in the bit allocation cache
b76ee706
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/10 11:42
Allocation cache can now be pre-computed as well.
5588d52e
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/10 11:07
Psychoacoustic decay coefficients can now be included in the static modes
defa3575
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/10 09:17
Simplified spreading function so that only one set of coefficients (the right
f39e869c
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/10 08:13
Initial support for static modes (defined STATIC_MODES)
2b361230
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/10 06:41
fixed-point: done converting find_spectral_pitch()
7b5dc851
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/09 13:18
manual stack allocator should now work for sizeof(char)!=1
2a6039b3
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/07 17:25
Adding code to dump the contents of a mode struct to a C file that can be
c209c342
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/07 12:50
fixed-point: defined HALF32() and used it for the forward mdct.
957a4fae
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/07 12:46
fixed-point: real FFT no longer needs float ops when MIXED_PRECISION is enabled.
3e650975
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/07 12:38
Bit of cleaning up. No real code change (well, I hope so!).
03892c18
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/07 12:25
fixed-point: finished intra_prediction(). No float ops left in vq.c
9455d1b1
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/07 12:17
fixed-point: converting intra-frame predictor (checkpoint #1).
000973f7
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/07 11:42
fixed-point: compute_band_energies() converted. Had to add some tiny bias to
0668b8ea
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/07 10:33
fixed-point: converted normalise_bands (had to split implementation)
cefc5ca2
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/07 09:53
Revert "compute_band_energies() merged with normalised_bands()"
f8eb420a
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/07 09:44
Changed the pulse spreading rotations so that the number of iterations is
ba8e2f7b
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/06 19:21
fixed-point: done converting quant_bands() and unquant_bands()
c4a40c65
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/05 18:22
doing spreading function and gain quantisation in-place
9d312b97
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/05 12:34
Defining RADIX_TWO_ONLY removes all butterflies, except for radix 2 and 4.
f93747c4
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/05 12:20
fixed-point: changed find_spectral_pitch() to use single-precision (16-bit) FFT.
b624cdd2
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/05 08:40
fixed-point: masking curve computation now converted. ***Fixed a bug in the
b0c93c66
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/05 08:12
oops, pgain_table now has half the number of elements
92376697
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/05 06:31
fixed-point: converted the pitch gain quantisation, except for the codebook
a1bc18a3
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/04 17:52
compute_band_energies() merged with normalised_bands()
069c3de6
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/04 15:38
Removed deprecated mode interface and added missing include
7137c26e
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/04 12:43
fixed-point: exp_rotation() now fully converted, using an approximation of the
a31899f0
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/03 17:28
added suffix to real fft. The celtclient can now be built with an unmodified
300cb5b8
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/03 12:52
First attempt at adding a suffix to kiss_fft functions so we can have multiple
bc60f8b3
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/03 12:23
making sure __GNUC_PREREQ() is only used when __GNUC__ is defined
730c6e39
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/03 11:38
Applying the window properly in find_spectral_pitch()
ddb181b7
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/03 09:53
A bit more reordering in stack memory allocation -- saved 1024 bytes on peak use
137ec8e9
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/03 09:44
applying the pitch windowing directly in find_spectral_pitch()
05080b4b
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/03 09:23
No longer storing the zero parts of the in[] array in the encoder
632f218b
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/03 09:08
Pitch predictor now uses a larger range of offset by not being limited to the
0141723d
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/03 08:59
compute_mdcts() no longer reads the part of the input that should be zeros
3e08a883
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/03 08:49
Only the (rising) overlap part of the window is stored now. No need to hold the
f28062fe
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/03 08:24
Don't bother multiplying when the window's value is one or zero
4c2787a4
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/03 05:04
Laplace encoder now works with 16-bit frequencies
37942649
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/01 17:55
Saved 4 kB of stack usage in find_spectral_pitch() by doing the FFT in-place
a556694e
– Jean-Marc Valin <
[email protected]>
authored
on 2008/03/01 15:14
Stereo pitch search no longer requires twice the memory/complexity.
Next