. "Nen\u00ED k dispozici"@cs . . . "21230" . . . "Sm\u00EDtka, Ji\u0159\u00ED" . "Nen\u00ED k dispozici"@cs . "RIV/68407700:21230/04:03101192" . "RIV/68407700:21230/04:03101192!RIV/2005/MSM/212305/N" . "Construction of Fully-Connected Hidden Markov Models of Environmental Sounds"@en . "Nen\u00ED k dispozici"@cs . "Sm\u00EDtkov\u00E1 Jank\u016F, Ladislava" . . . "Construction of Fully-Connected Hidden Markov Models of Environmental Sounds" . . . "[F13E828015A2]" . . "Z(MSM 212300014)" . "Classification of General Audio; Environmental Sounds; Hidden Markov Models; Pattern Recognition"@en . "2"^^ . . "3-901608-28-1" . . . "Construction of Fully-Connected Hidden Markov Models of Environmental Sounds"@en . "Vienna" . "2"^^ . "558644" . "This paper deals with the estimation of parameters of fully-connected continuous Hidden Markov Models of environmental sounds. A method based on combining K-variable K-means algorithm and EM algorithm has been used for estimation of number of states of Hidden Markov chain. State likelihoods has been modelled via triangular mixtures. Constructed HMMs have been tested on a database of real-world environmental sounds. A classification accuracy of 89 percent has been reported."@en . "Construction of Fully-Connected Hidden Markov Models of Environmental Sounds" . "This paper deals with the estimation of parameters of fully-connected continuous Hidden Markov Models of environmental sounds. A method based on combining K-variable K-means algorithm and EM algorithm has been used for estimation of number of states of Hidden Markov chain. State likelihoods has been modelled via triangular mixtures. Constructed HMMs have been tested on a database of real-world environmental sounds. A classification accuracy of 89 percent has been reported." . . "neuvedeno" . . . . .