- Dec 20, 2023
-
-
Jean-Marc Valin authored
Need to move the redundant frame even in CBR because the hybrid frame now gets encoded as VBR, with DRED picking up the rest. Fixes an issue introduced in 4600e775.
-
Jean-Marc Valin authored
The encoder wouldn't reserve enough bits for CELT, causing it to not have enough bits to code the switching redundancy flag when it should have.
-
-
- Dec 17, 2023
-
-
Jean-Marc Valin authored
Avoids valgrind complaining about use of uninitialized memory
-
- Dec 15, 2023
-
-
Michael Klingbeil authored
-
- Dec 14, 2023
-
-
Michael Klingbeil authored
-
- Dec 06, 2023
-
-
Michael Klingbeil authored
-
- Dec 05, 2023
-
-
Jean-Marc Valin authored
-
- Nov 30, 2023
-
-
Michael Klingbeil authored
-
Michael Klingbeil authored
-
- Nov 29, 2023
-
-
Jean-Marc Valin authored
-
Michael Klingbeil authored
-
Jean-Marc Valin authored
Removes float debug weights, as well as useless spaces
-
Jean-Marc Valin authored
-
Jean-Marc Valin authored
-
Jean-Marc Valin authored
Use OPUS_DRED instead of NEURAL_FEC
-
Jean-Marc Valin authored
Also default to disabling dnn float debugging
-
Jean-Marc Valin authored
Still don't quite know what I'm doing
-
- Nov 28, 2023
-
-
Jean-Marc Valin authored
-
Jean-Marc Valin authored
Compilers sometimes replace vmlaq*() with fmul+fadd instead of fmla. Trying to use vfmaq*() instead when possible.
-
Jean-Marc Valin authored
Finished adversarial training on 800k model. Also, move weights to a new location.
-
Jean-Marc Valin authored
1) Enable asm/intrinsics even for floating-point 2) Make sure ARMv8 asimd enables EDSP/MEDIA/Neon 3) Add dotp architecture to rtcd table since AArch *can* have dotp
-
- Nov 27, 2023
-
-
Jean-Marc Valin authored
Adds RTCD tables for compute_activation() and compute_conv2d()
-
Jean-Marc Valin authored
-
Jean-Marc Valin authored
Avoids having to write intrinsics for simple loops
-
Jean-Marc Valin authored
Enabling only on platforms that have been tested just in case we run into a non-IEEE754 platform where they would break.
-
Jean-Marc Valin authored
-
- Nov 26, 2023
-
-
Jean-Marc Valin authored
Still missing some intrinsics
-
Jean-Marc Valin authored
-
- Nov 25, 2023
-
-
Jean-Marc Valin authored
Used for DNN matrix multiplies
-
- Nov 24, 2023
-
-
Jean-Marc Valin authored
-
Jean-Marc Valin authored
-
Jean-Marc Valin authored
-
Jean-Marc Valin authored
Code moved to compute_frame_features()
-
Jean-Marc Valin authored
-
- Nov 21, 2023
-
-
Jean-Marc Valin authored
for cmake, force PRESEUME_SSE4_1 on PRESUME_AVX2
-
Jean-Marc Valin authored
aka banging on it until it builds on my machine. Further improvements welcome
-
Jean-Marc Valin authored
-
Jean-Marc Valin authored
Not yet with rtcd
-
Jean-Marc Valin authored
-