Kaldi is a very useful yet scarcely documented tool for ASR ( Automatic Speech Recognition).
This blog is an attempt towards sharing the installation instructions for this tool on ubuntu.
Steps to Install kaldi:
Step 1: Goto your root directory
Step 2: Clone kaldi from git using :
git clone https://github.com/kaldi-asr/kaldi.git kaldi --origin upstream
Step 3: Goto the kaldi folder from root using :
cd kaldi
Step 4: Get the updates in the folder ( optional step )
git pull
Step 5: Goto tools/extras folder within kaldi and check for dependencies if any using:
cd kaldi/tools/extras
./check_dependencies.sh
Step 6: In case of any dependencies make those installations and repeat Step 5 ( Note: all the depedencies are to be installed in the tools folder)
Continue this till all the dependencies are installed.
Step 7 : Goto the tools folder ( kaldi/tools) and install ATLAS headers, openFst,SCTK and sph2pipe.
Use Command:
make
or
make CXX=g++-4.8 (to specify the exact version of compiler)
or
make -j 4 ( to use 4 cores in parallel) (process completes fast)
Step 8: Goto src folder in kaldi ( kaldi/src) Use command:
cd .. ( to exit the tools folder and return back to the kaldi)
cd src ( to enter teh src folder)
Step 9: Final Installation step :
Use commands:
./configure --shared
make depend -j 8
make -j 8
Note : Instead of -j 8 we can use -j “x” where is the total count of cores available in our machine.
Upon executing this the Installation is complete.
Thanks for Reading!
Stay tuned for further blogs on related topics!