 Tristan Matthews
|
39a7979ad7
cosmetics: drop trailing whitespace
|
6 miesięcy temu |
|
Tristan Matthews
|
c1f82d2148
math_approx: use unsigned int for LCG pseudorandom generator
|
5 lat temu |
 Ron
|
e2d3bc3923
Tidy the speex/*.h includes some more
|
11 lat temu |
 John Ridges
|
73dee02261
Removing h1_mem from the encoder state since it wasn't used at all.
|
15 lat temu |
 Jean-Marc Valin
|
785e61a593
Sorting out #include "" from #include <> and using relative paths for including
|
16 lat temu |
 jm
|
f39618b8d4
New DISABLE_VBR and DISABLE_FLOAT_API switches can together remove every single
|
17 lat temu |
|
jm
|
84064b108f
Lag window now stored as const data instead of computed at init time. Also,
|
17 lat temu |
|
jm
|
1a37315f81
sb_celp: field bufSize was no longer used, so it's now removed
|
17 lat temu |
|
jm
|
af21325a16
Patch by Thom Johansen: change memory of qmf_synth to 16-bit.
|
17 lat temu |
|
jm
|
a717314ed5
Moved filters from state to stack for wideband encoder (~180 bytes in fixed-
|
17 lat temu |
|
jm
|
bbb1ed56c7
Moved 20 filter coefs from state to stack in wideband decoder.
|
17 lat temu |
|
jm
|
08151da50f
Only the RMS of the excitation is saved in the encoder -- 320 bytes off the
|
18 lat temu |
|
jm
|
809a4507f4
Most of the wideband encoder now in 16-bit. Also, saved another 1.5 kB off the
|
18 lat temu |
|
jm
|
a7cb0d33d4
zero-response now in 16-bit and stored on the stack. About 1 kB saved off the
|
18 lat temu |
|
jm
|
8f2809edc8
No need for an explicit buffer to receive the narrowband innovation.
|
18 lat temu |
|
jm
|
71d983adf9
Removed 5 kB of useless buffering in the wideband encoder. Also, fixed the
|
18 lat temu |
|
jm
|
678e1bd9ce
Another 1 kB off the encoder state by converting the qmf_decomp output to
|
18 lat temu |
|
jm
|
32accf86c4
Got rid of the excitation memory in the decoder and changed the relevant vars
|
18 lat temu |
|
jm
|
b544a9278b
Saved innovation from nb to wb is now 16-bit precision.
|
18 lat temu |
|
jm
|
c679c22021
another 640 bytes off the decoder state.
|
18 lat temu |
|
jm
|
29dba66464
Saved another 2kB from the decoder by converting high-band synthesis to
|
18 lat temu |
|
jm
|
a629adae51
Replaced fir_mem_up by qmf_synth which doesn't require intermediate
|
18 lat temu |
|
jm
|
f4f59e83d0
copyright details, more info about the mdf learning rate
|
18 lat temu |
|
jm
|
bda20177a4
Max VBR bit-rate implemented for wideband (semi tested), still doesn't quite
|
18 lat temu |
|
jm
|
e2d8a7745d
changed a few parameter definitions from int to spx_int32_t. Should by more
|
18 lat temu |
|
jm
|
28ac6c6418
SPEEX_SET_VBR_MAX_BITRATE partially implemented for wideband (still doesn't
|
18 lat temu |
|
jm
|
1f16e63407
excitation and target are now 16-bit (were 32) in the encoder
|
18 lat temu |
|
jm
|
8f6aaa037d
fixed enhancement for wideband (sort of) and a bit of cleaning up.
|
18 lat temu |
|
jm
|
4fc6573ebc
innovation is saved directly to the sb_celp en/decoder, so a buffer is
|
18 lat temu |
|
jm
|
7b51de5b4f
LPC analysis window now stored as constant in code -> 400 bytes off the
|
18 lat temu |
