Viterbi

From lingwiki

Revision as of 18:21, 5 September 2009 (edit) Abney (Talk | contribs) ← Older edit

Current revision (13:04, 11 November 2009) (edit) (undo) Abney (Talk | contribs) m (1 revision)

---

Current revision

The Viterbi Algorithm is an algorithm which takes a particular HMM, and from an output that has been observed, determines the most likely sequence of states which caused that output.

[edit] Use in Speech Recognition

This algorithm is highly used in determining state sequences of HMMs in the acoustic model and language model of a speech recognizer.