# PyKaldi whl packages for Linux/Ubuntu These whl packages make installing pykaldi much easier on your system. We currently have packages for Python 3.6, 3.7 and 3.8. Note that you need to compile Kaldi separately, its not sufficient to just install the whl package. See the installing notes below. # Installing and using the packages Make sure you have the Python dev package and the correct Python version installed: sudo apt-get install python3.8 python3.8-dev You can also follow this installation guide with Python 3.7 or 3.6, if 3.8 is not available on your system. Create a new project folder, for example: mkdir -p ~/projects/myASR cd ~/projects/myASR Download the whl package from https://ltdata1.informatik.uni-hamburg.de/pykaldi/ and all shell scripts to this folder: wget https://ltdata1.informatik.uni-hamburg.de/pykaldi/pykaldi-0.1.2-cp38-cp38-linux_x86_64.whl wget https://ltdata1.informatik.uni-hamburg.de/pykaldi/install_kaldi.sh wget https://ltdata1.informatik.uni-hamburg.de/pykaldi/install_kaldi_intel.sh wget https://ltdata1.informatik.uni-hamburg.de/pykaldi/path.sh wget https://ltdata1.informatik.uni-hamburg.de/pykaldi/install_mkl.sh chmod guo+x *.sh Install virtualenv: sudo python3.8 -m pip install virtualenv In your project folder, create a virtual environment with Python 3.8 and activate it: virtualenv -p /usr/bin/python3.8 pykaldi_env . pykaldi_env/bin/activate Install the pykaldi package: pip3 install pykaldi-0.1.2-cp38-cp38-linux_x86_64.whl You need to compile Kaldi as well, since it is not included in the whl package: Install Kaldi and Intel MKL if you have an Intel CPU (see note below if you have a different CPU than Intel) ./install_mkl.sh ./install_kaldi_intel.sh For a non-Intel CPU like AMD do this instead: ./install_kaldi.sh If no compile errors occured, then PyKaldi is ready to be used in your project! Source Kaldi paths, so pykaldi can find everything: . path.sh Now you should be able to import kaldi in Python and use it. If you want to use your project in a fresh shell session, you'd need to run: . pykaldi_env/bin/activate . path.sh To load the environment and source the required paths into your shell environment.