 DataHoarder
|
a2e2148b78
|
Support decoding into Float
|
2022-10-04 10:43:29 +02:00 |
|
Sean DuBois
|
27b0cc9a06
|
Fix README for examples/playback
URL to faiface was poorly formatted
|
2022-10-03 23:17:57 -04:00 |
|
|
Sean DuBois
|
a9917652d9
|
Add blogpost to README
|
2022-10-03 00:03:54 -04:00 |
|
|
Sean DuBois
|
2d6fc47384
|
Panic on decode error in examples/playback
Error wasn't returned to user before
|
2022-10-02 21:11:11 -04:00 |
|
|
Sean DuBois
|
5777c86e20
|
Update README
|
2022-10-01 23:37:17 -04:00 |
|
|
Sean DuBois
|
dbc1ce8df2
|
Add examples/playback
Using faiface/beep play a Ogg file directly from disk
|
2022-10-01 23:09:51 -04:00 |
|
|
Sean DuBois
|
733188ab17
|
Implement bitdepth conversion and upsampling
bitdepth package takes F32LE -> S16LE and handles upsampling
from 16Khz -> 48Khz
|
2022-09-30 21:56:49 -06:00 |
|
|
Sean DuBois
|
7941dd8c63
|
Finish LTP Synthesis Implementation
Silk Decoder is fully implemented!
|
2022-09-30 16:08:38 -07:00 |
|
|
Sean DuBois
|
684adce144
|
Use LPC values from previous when current frame has no cached values
|
2022-09-25 22:42:19 -04:00 |
|
|
Sean DuBois
|
238ccbfc34
|
Properly implement Q15 handling
If w4 == 4 the normalized value (n1Q15) is only used for the first two subframes.
Before we incorrectly would use it for all subframes.
|
2022-09-25 22:29:01 -04:00 |
|
|
Sean DuBois
|
4c632df5ac
|
Remove finalLPCValues from silk.Decoder
LPC values shouldn't be stored across frames
|
2022-09-25 20:22:17 -04:00 |
|
|
Sean DuBois
|
3a79dec65f
|
Create cache for res and out values
LTP Synthesis needs values of out + res for previous frames
|
2022-09-22 23:32:16 -04:00 |
|
|
Sean DuBois
|
903249e356
|
Don't recompute res for each subframe
|
2022-09-21 23:02:46 -04:00 |
|
|
Sean DuBois
|
42288fd1e3
|
Fix LTP Scaling Parameter generation
Was incorrectly not subframe aware
|
2022-09-21 16:08:50 -04:00 |
|
|
Sean DuBois
|
9886d0ca79
|
Fix decode README
ffmpeg + GStreamer commands were incorrect
|
2022-09-21 13:42:30 -04:00 |
|
|
Sean DuBois
|
eb2f85b871
|
Fix function name
floatarrtobytes -> convertFloatToByteSlice
|
2022-09-20 23:38:23 -04:00 |
|
|
Sean DuBois
|
1197e6fe64
|
Improve README
|
2022-09-20 23:30:26 -04:00 |
|
|
Sean DuBois
|
59ab961a93
|
Start READMEs
|
2022-09-20 23:09:04 -04:00 |
|
|
Sean DuBois
|
38e75cb5cf
|
Upgrade oggreader to be segment aware
|
2022-09-20 22:29:27 -04:00 |
|
|
Sean DuBois
|
7b32fb7f78
|
Properly pass decoded excitation into LTP Synthesis
Before would discard excitation and always use a zero'ed out list
|
2022-09-20 15:58:02 -04:00 |
|
|
Sean DuBois
|
9e56e15906
|
Use float32 instead of float64 internally
|
2022-09-20 14:17:13 -04:00 |
|
|
Sean DuBois
|
c631824345
|
Add untested LTP Synthesis implementation
Implements the spec, but haven't compared to values from libopus yet
|
2022-09-18 23:16:42 -04:00 |
|
|
Sean DuBois
|
cb995e5d2b
|
Update PitchLag decoder to be subframe aware
Before would only compute s=0
|
2022-09-17 16:20:53 -04:00 |
|
|
Sean DuBois
|
c9124c5c4e
|
Move Silk Frame Reconstruction into dedicated method
More closely matches RFC
|
2022-09-16 22:45:37 -04:00 |
|
|
Sean DuBois
|
0112ac2b3e
|
Make more state available to LTP Synthesis
Unimplemented but needed
|
2022-09-16 22:32:17 -04:00 |
|
Sean DuBois
|
56e47b2485
|
Update lpcSynthesis to take subframe only of out
Instead of doing arithmetic in the function make it the responsibility
of the caller.
|
2022-09-16 12:45:45 -04:00 |
|
|
Sean DuBois
|
ce891b494b
|
Fix rollover issue with logGain
Was incorrectly stored as unsigned integer before
|
2022-09-15 21:55:19 -04:00 |
|
|
Sean DuBois
|
782c09fe4c
|
Fix LPC Synthesis Bug
Use gainQ16 for the current subframe. Before we always used gainQ16[0]
incorrectly.
|
2022-09-15 15:03:33 -04:00 |
|
|
Sean DuBois
|
2e21fbab86
|
Add full decode test
Test a full unvoiced frame through Silk decoder
|
2022-09-15 13:40:42 -04:00 |
|
|
Sean DuBois
|
5d12791c86
|
Add test cases for all subframes
LPC unit test was only checking first run
|
2022-09-13 15:59:05 -04:00 |
|
|
Sean DuBois
|
b9266e005e
|
Improve constant naming
add `codebook` prefix to vars that needed it
|
2022-09-13 11:49:59 -04:00 |
|
|
Sean DuBois
|
e2d581a590
|
Finish decodeLTPScalingParamater
Defined in section-4.2.7.6.3
|
2022-09-12 22:08:56 -04:00 |
|
|
Sean DuBois
|
2067ca5ac0
|
Implement decoding LTP Filter Coefficients
section-4.2.7.6.2
|
2022-09-12 17:14:13 -04:00 |
|
Sean DuBois
|
1dcc971611
|
Delete unused function
|
2022-09-11 23:33:50 -04:00 |
|
|
Sean DuBois
|
87d5d99155
|
Implement Pitch Lags
Defined in section-4.2.7.6.1
|
2022-09-11 23:32:29 -04:00 |
|
|
Sean DuBois
|
85ca7b30c0
|
Fix frameType parsing
frameTypeSymbol values start from 0 again
when voiceActivityDetected == true
|
2022-09-08 00:41:09 -04:00 |
|
|
Sean DuBois
|
e005fba189
|
Finish normalizeLSFInterpolation
Before we only handled when wQ2 == 4
|
2022-09-08 00:16:35 -04:00 |
|
|
Sean DuBois
|
496497d1f7
|
Add test for LPC Synthesis
|
2022-09-07 16:01:09 -04:00 |
|
|
Sean DuBois
|
e06e1cdfef
|
Add test for Silk limitLPCFilterPredictionGain
|
2022-09-07 11:48:02 -04:00 |
|
|
Sean DuBois
|
b2df0354df
|
Implement LPC Synthesis
Some data makes it all the way through. LTP Synthesis needs to be
implemented now
|
2022-09-06 23:49:51 -04:00 |
|
|
Sean DuBois
|
f20a9eedf1
|
Add stubs for rest of Silk Decoder
Also include the implemented excitation functions now that we have
reached them.
|
2022-09-05 23:50:07 -04:00 |
|
|
Sean DuBois
|
5919d29d78
|
Add stub for Limiting the Prediction Gain of the LPC Filter
Comes from 4.2.7.5.8 we can implement this later
|
2022-09-05 15:41:26 -04:00 |
|
|
Sean DuBois
|
1e8a69eba3
|
Implement Limiting the Prediction Gain of the LPC Filter
Don't have a test to actually exercise this behavior yet. All the test
files I have currently don't enter this block.
Defined in section-4.2.7.5.7
|
2022-09-03 23:54:52 -04:00 |
|
|
Sean DuBois
|
733dc27100
|
Implement Converting Normalized LSFs to LPC Coefficients
From 4.2.7.5.6
|
2022-08-29 22:16:56 -04:00 |
|
|
Sean DuBois
|
b5ea7c4523
|
Finish decoder Excitation
Implements final part
https://datatracker.ietf.org/doc/html/rfc6716#section-4.2.7.8.6
|
2022-08-10 01:15:26 -04:00 |
|
|
Sean DuBois
|
8412cf5190
|
Split SILK excitation implementation into functions
Easier to follow and matches established pattern
|
2022-08-10 00:18:46 -04:00 |
|
|
Sean DuBois
|
cfae85659a
|
Implement Excitation Sign Decoding
Specified in https://datatracker.ietf.org/doc/html/rfc6716#section-4.2.7.8.5
|
2022-08-08 23:50:53 -04:00 |
|
|
Sean DuBois
|
dd17102d36
|
Implement Pulse Location Decoding
Defined at https://datatracker.ietf.org/doc/html/rfc6716#section-4.2.7.8.3
|
2022-08-08 00:56:59 -04:00 |
|
|
Sean DuBois
|
65e49e6622
|
Add all Pulse Count ICDFs
From https://datatracker.ietf.org/doc/html/rfc6716#section-4.2.7.8.3
|
2022-08-07 14:46:21 -04:00 |
|
|
Sean DuBois
|
a7a80f67e8
|
Fix incorrect value in test
Used wrong value from libopus RangeDecoder when comparing values
|
2022-08-07 00:19:07 -04:00 |
|