This package contains sample data and scripts for creation of Czech spontaneous speech recogniser. Within the training on limited amount of data, simple language model is created as well as models of Czech monophones and triphones.