The problem of speech recognition is one of signal processing, model design, equivalence classification (clustering), parameter estimation from data, and hypothesis search.

The field therefore impinges in many of its aspects on information theory. The talk will present some solutions as well as challenges.