- To run the executable open a Command Prompt in Windows and run the script using the following syntax:
classify.exe -i <input_map_name> -o <output_map_name> -s <sampling_rate> -m <use case> [optional] -c <channels> -b <bit_depth>
- The use case option has to be one of the following
- MEC, ASC, PMLDC, CGDLC, ITC
Example usage
classify.exe -i input -o output -s 16000 -m MEC
-
The sampling rate is used to read the input_map bytes. The default sound bit depth and sampling rate is 32 and 16000, meaning that by default we expect to read 64000 bytes from the input map.
-
To create the .exe the pyinstaller package was used with the following syntax:
pyinstaller.exe --onefile classify.spec
To use the .spec file you have to download the script dependencies and change the paths according to your own system. The libraries needed were downloaded using the pip tool.
pip install scipy
pip install numpy
pip install keras
pip install librosa
pip install cntk
The cntk package was used as an alternative backend library to tensorflow to read the Neural Net model.
To use cntk as the keras backend
set KERAS_BACKEND=cntk