Kaldi is an open source toolkit for speech recognition, intended for use by speech recognition researchers and professionals. It was developed initially at Johns Hopkins University with contributions ...