Matches in SemOpenAlex for { <https://semopenalex.org/work/W2093658107> ?p ?o ?g. }
Showing items 1 to 76 of
76
with 100 items per page.
- W2093658107 endingPage "410" @default.
- W2093658107 startingPage "397" @default.
- W2093658107 abstract "HMM-based automatic segmentation has been popularly used for corpus construction for concatenative speech synthesis. Since the most important reasons for the inaccuracy of HMM-based automatic segmentation are the HMM training criterion and duration control, we will study these particular issues. For the HMM training, we apply the discriminative training method and introduce a new criterion, named Minimum SeGmentation Error (MSGE). In this method, a loss function directly related to the segmentation error is defined, and parameter optimization is performed by the Generalized Probabilistic Descent (GPD) algorithm. For the duration control problem, we apply explicit duration models and propose a two-step-based segmentation method to solve the problem of computational cost, where the duration model is incorporated in a postprocessor procedure. From the experimental results, these two techniques significantly improve segmentation accuracy with different focuses, where the MSGE-based discriminative training focuses on improving the accuracy of sensitive boundary, i.e., a boundary where an error in segmentation is likely to cause a noticeable degradation in speech synthesis quality, and the explicit duration modeling focuses on eliminating large errors. After combining these two techniques, the error average was reduced from 6.86 ms to 5.79 ms on Japanese data, and from 8.67 ms to 6.61 ms on Chinese data. Simultaneously, the number of errors larger than 30 ms were reduced 25% and 51% on Chinese and Japanese data, respectively." @default.
- W2093658107 created "2016-06-24" @default.
- W2093658107 creator A5004733344 @default.
- W2093658107 creator A5025903844 @default.
- W2093658107 creator A5039492028 @default.
- W2093658107 creator A5069927701 @default.
- W2093658107 date "2005-12-01" @default.
- W2093658107 modified "2023-10-18" @default.
- W2093658107 title "Discriminative training and explicit duration modeling for HMM-based automatic segmentation" @default.
- W2093658107 cites W114226241 @default.
- W2093658107 cites W146439846 @default.
- W2093658107 cites W1481105745 @default.
- W2093658107 cites W1594082249 @default.
- W2093658107 cites W168484277 @default.
- W2093658107 cites W1979761486 @default.
- W2093658107 cites W2064218608 @default.
- W2093658107 cites W2100295236 @default.
- W2093658107 cites W2105080323 @default.
- W2093658107 cites W2115678687 @default.
- W2093658107 cites W2125838338 @default.
- W2093658107 cites W2149546853 @default.
- W2093658107 cites W2158289097 @default.
- W2093658107 cites W4302326871 @default.
- W2093658107 doi "https://doi.org/10.1016/j.specom.2005.03.016" @default.
- W2093658107 hasPublicationYear "2005" @default.
- W2093658107 type Work @default.
- W2093658107 sameAs 2093658107 @default.
- W2093658107 citedByCount "5" @default.
- W2093658107 countsByYear W20936581072014 @default.
- W2093658107 countsByYear W20936581072021 @default.
- W2093658107 crossrefType "journal-article" @default.
- W2093658107 hasAuthorship W2093658107A5004733344 @default.
- W2093658107 hasAuthorship W2093658107A5025903844 @default.
- W2093658107 hasAuthorship W2093658107A5039492028 @default.
- W2093658107 hasAuthorship W2093658107A5069927701 @default.
- W2093658107 hasConcept C112758219 @default.
- W2093658107 hasConcept C124952713 @default.
- W2093658107 hasConcept C142362112 @default.
- W2093658107 hasConcept C153180895 @default.
- W2093658107 hasConcept C154945302 @default.
- W2093658107 hasConcept C23224414 @default.
- W2093658107 hasConcept C28490314 @default.
- W2093658107 hasConcept C41008148 @default.
- W2093658107 hasConcept C89600930 @default.
- W2093658107 hasConcept C97931131 @default.
- W2093658107 hasConceptScore W2093658107C112758219 @default.
- W2093658107 hasConceptScore W2093658107C124952713 @default.
- W2093658107 hasConceptScore W2093658107C142362112 @default.
- W2093658107 hasConceptScore W2093658107C153180895 @default.
- W2093658107 hasConceptScore W2093658107C154945302 @default.
- W2093658107 hasConceptScore W2093658107C23224414 @default.
- W2093658107 hasConceptScore W2093658107C28490314 @default.
- W2093658107 hasConceptScore W2093658107C41008148 @default.
- W2093658107 hasConceptScore W2093658107C89600930 @default.
- W2093658107 hasConceptScore W2093658107C97931131 @default.
- W2093658107 hasIssue "4" @default.
- W2093658107 hasLocation W20936581071 @default.
- W2093658107 hasOpenAccess W2093658107 @default.
- W2093658107 hasPrimaryLocation W20936581071 @default.
- W2093658107 hasRelatedWork W1507687735 @default.
- W2093658107 hasRelatedWork W1565453839 @default.
- W2093658107 hasRelatedWork W1652783584 @default.
- W2093658107 hasRelatedWork W198264392 @default.
- W2093658107 hasRelatedWork W2024160000 @default.
- W2093658107 hasRelatedWork W2134999511 @default.
- W2093658107 hasRelatedWork W2167155152 @default.
- W2093658107 hasRelatedWork W2355170890 @default.
- W2093658107 hasRelatedWork W2404514746 @default.
- W2093658107 hasRelatedWork W2510758617 @default.
- W2093658107 hasVolume "47" @default.
- W2093658107 isParatext "false" @default.
- W2093658107 isRetracted "false" @default.
- W2093658107 magId "2093658107" @default.
- W2093658107 workType "article" @default.