Skip to content
Snippets Groups Projects
Select Git revision
  • deinterleaved_channels
  • main default protected
  • CI_test
3 results
Blame Permalink
Forked from vceg / bwc
20 commits behind the upstream repository.
README.md 2.57 KiB

Versatile Adaptive Waveform Codec (VAWC)

The VAWC code is located in directory source/. Executable targets are EncoderApp and DecoderApp.

build codec (linux)

  • git clone
  • select/create branch
  • create build dir (if not present)
mkdir build
  • generate project files via cmake (cmake needs to be installed) OR use ide like vscode (with cmake extension)
cmake -S . -B build -DCMAKE_BUILD_TYPE=Release
  • build project
cmake --build build --config Release --target all
  • binaries are copied to bin/folder after building -- xxxAppStatic (Release) -- xxxAppStaticd (Debug)

build codec (windows)

  • git clone

  • select/create branch

  • create build dir (if not present)

  • generate project files via cmake (developer shell e.g., cmake needs to be installed) OR use ide like vscode (with cmake extension)

cmake -S . -B build
  • build project from command line (developer shell) OR use ide like vscode
cmake --build build --config Release --target ALL_BUILD

config files and data categories

  • currently, 3 different categories of data are used: ECG, EMG and EEG. For ECG, it is recommended to use the config file combinedPresetECG.cfg, for EEG it is recommended to use the config file combinedPresetEEG.cfg, and for EMG it is recommended to use the config file combinedPresetEMG.cfg.

run Codec Apps Examples

  • run encoder
<encoder_executable> --InputFile="input" --InputBitDepth="bd" -c combinedPresetXXX.cfg --BitstreamFile="bitstream" --FileFormat=16bitSignedHeader

where "input" is the path to the input file, "bistream" is the path to the bitstream file to be generated and "bd" is the input bit depth, which is 12 for the ECG dataset and 16 for the EMG and EEG datasets that were used in the CfP. The input file can be either an EDF-file, a WAV-file or a dat-file. Here, dat-files are organized as follows: The first two bytes determine the number of input channels as 16 bit unsigned integer. The remaining bytes are the sample values as 16 bit signed integers in an interleaved format, i.e. ch0-s0, ch1-s0,...,chM-s0, ch0-s1, ch1-s1,..., chM-s1,...where chi denotes the i-th channel and sj denotes the j-th time index and where M is the number of channels.

  • run decoder
<decoder_executable>  --BitstreamFile="bitstream" --OutputFile="output" --FileFormat=16bitSignedHeader

where "bitstream" is the path to the bitstream file to be decoded and "output" is the path of the reconstructed waveform file to be generated by the decoding process. The latter file is always generated as a dat file defined as above.