statistical language model training description
  • aim

To provide a general level of knowledge and gain pratical competencies in building statistical language model used in continuous speech recognition.

  • duration: 2 days
  • audience
  • voice applications developers
  • project managers
  • pre requisites
  • grammar development and optimisation training
  • project life cycle and quality control
  • speech technology, in particular speech recognition
  • script language programming
  • topics

The training covers the following topics:

  • theory of statistical language model (SLM)
  • SLM design process
  • SLM grammar compilation
  • the range and use of the teliSpeech SLM toolkit
  • the improvement of SLM recognition performance within the general tuning process.
  • delivery method: presentations and practical sessions
  • languages: English, French