Method, apparatus, and radio for optimizing hidden markov model speech recognition
CA 2189249 C · Owner: Srenger, Edward · Inventors: Srenger, Edward; Kushner, William M.; Hartman, Matthew A.
Overview
Its claims, description and drawings are in the Canadian Patents Database.
Open the CIPO recordAbstract From the Canadian patent
In a statistical based speech recognition system one of the, key issues is the selection of the Hidden Markov Model that best matches a given sequence of feature observations. The problem is usually addressed by the calculation of the maximum likelihood ML, state sequence by means of a Viterbi or other decoder. Noise or inadequate training can produce an ML sequence associated with a Hidden Markov Model other than the correct model. The method of the present invention provides improved robustness by combining the standard ML state sequence score (416) with an additional path core (418) derived from the dynamics of the ML score as a function of time. These two scores, when combined, form a …
Read the whole patent
Listed from its CIPO record in Patent Yard's archive of Canadian patents past their term. Its claims, description and drawings are in the CIPO record and on Google Patents, and the Claude kit works from them.
Why it's free to use
- Filed January 29, 1996 and granted April 10, 2001. A Canadian patent filed from October 1, 1989 lasts 20 years from filing, so its term ended January 29, 2016.
- A certificate of supplementary protection adds at most two years for a drug, so even that would have ended by January 29, 2018.
- We check Canadian rights only. A patent in its family in another country could still be in force: check its family on Google Patents.
Timeline
Timeline From CIPO dates
Sources & verification
Official CIPO records
Everything on this page comes from the CIPO record linked above. Contains information licensed under the Open Government Licence – Canada.