FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2170938943> ?p ?o ?g. }

• W2170938943 abstract "The prosodic structure of utterances in terms of breaks and tones is a significant problem in speech synthesis. In this work we present the results from models used to predict accurate and realistic prosodic structures within the context of a Concept-to-Speech system for a virtual museum guide. We have used a Natural Language Generator system for providing error-free enriched linguistic information, such as syntax and Part-of-Speech, to a Speech Synthesizer. An XML annotation has been used as a means for this transfer of linguistic data. The annotated data was used to build classification trees for the prediction of prosodic phrase breaks, pitch accents and endtones (phrase accents and boundary tones). The annotation of utterances included segmental information, ToBI marks, syntax, grammar and some domain specific features such as new/given and phrase subject/object information. The linguistic nature of the domain allowed us to carefully select the set of features and the training conditions and also to utilize speech-oriented information from the written language produced by the Natural Language Generator component, such as evidence of stress and intonational focus. A speech corpus of 516 utterances has been used for training and evaluation purposes. To optimize the generated models, we used exhaustive training upon the domain data, achieving a correlation between the observed and the predicted elements of 97.286% for phrase breaks, 99.349% for pitch accents and 99.992% for endtones. One of the most important tasks in Text-to-Speech (TtS) synthesis is the prediction of the prosodic structure of the utterance to be spoken. This forms the basis of the rendering of the segmental durations and the fundamental frequency's contour or the selection of the appropriate units in corpus based synthesis. The description of the prosodic structure is usually defined by the position and the type of (a) prosodic phrase breaks, (b) pitch accents, (c) phrase accents and (d) boundary tones. The last two are usually grouped together (they do not co-occur in the tone tier) and referred as endtones. Two approaches are usually followed for the identification of the above elements in an utterance: rule-driven and machine learning. The former fails to capture all the richness of human speech, is generally difficult to write and to adapt to new domains and usually provides poor input to the prosody generation module, while the latter can yield reasonable results as long as the size of the sample data increases with the size of the domain of the application. Prosody generation is a complex process that involves the analysis of several linguistic phenomena. Dynamic approaches are usually prone to errors. For instance, part-of- speech (POS) identification fails in 5% of the cases for Greek using statistical taggers (Petasis et al., 1999), while syntax and metric trees are hard to construct. A solution that overcomes that is offered by (a) limiting the domain to which the TtS applies to and thus limiting the linguistic phenomena, and (b) using a Concept-to-Speech (CtS) system (Theune et al., 2001). The advantage of the latter is that the generated texts are annotated with high level linguistic factors in contrast to plain texts (Reiter and Dale, 1997). Annotation varies in terms of" @default.
• W2170938943 created "2016-06-24" @default.
• W2170938943 creator A5004819616 @default.
• W2170938943 creator A5034227751 @default.
• W2170938943 creator A5050644786 @default.
• W2170938943 date "2003-01-01" @default.
• W2170938943 modified "2023-09-27" @default.
• W2170938943 title "Building Prosodic Structures in a Concept-to-Speech System" @default.
• W2170938943 cites W1521476198 @default.
• W2170938943 cites W1529629239 @default.
• W2170938943 cites W1546533840 @default.
• W2170938943 cites W155051124 @default.
• W2170938943 cites W1582784770 @default.
• W2170938943 cites W1594031697 @default.
• W2170938943 cites W1604793456 @default.
• W2170938943 cites W1621817313 @default.
• W2170938943 cites W169585292 @default.
• W2170938943 cites W1970354675 @default.
• W2170938943 cites W1976066153 @default.
• W2170938943 cites W2025617415 @default.
• W2170938943 cites W2055418239 @default.
• W2170938943 cites W2093476462 @default.
• W2170938943 cites W2114141960 @default.
• W2170938943 cites W2124924501 @default.
• W2170938943 cites W2125394509 @default.
• W2170938943 cites W2159465751 @default.
• W2170938943 cites W26367467 @default.
• W2170938943 cites W3085162807 @default.
• W2170938943 hasPublicationYear "2003" @default.
• W2170938943 type Work @default.
• W2170938943 sameAs 2170938943 @default.
• W2170938943 citedByCount "0" @default.
• W2170938943 crossrefType "journal-article" @default.
• W2170938943 hasAuthorship W2170938943A5004819616 @default.
• W2170938943 hasAuthorship W2170938943A5034227751 @default.
• W2170938943 hasAuthorship W2170938943A5050644786 @default.
• W2170938943 hasConcept C120665830 @default.
• W2170938943 hasConcept C121332964 @default.
• W2170938943 hasConcept C14999030 @default.
• W2170938943 hasConcept C154945302 @default.
• W2170938943 hasConcept C177264268 @default.
• W2170938943 hasConcept C192209626 @default.
• W2170938943 hasConcept C195324797 @default.
• W2170938943 hasConcept C199360897 @default.
• W2170938943 hasConcept C204321447 @default.
• W2170938943 hasConcept C2776224158 @default.
• W2170938943 hasConcept C28490314 @default.
• W2170938943 hasConcept C41008148 @default.
• W2170938943 hasConcept C53893814 @default.
• W2170938943 hasConcept C542774811 @default.
• W2170938943 hasConcept C60048249 @default.
• W2170938943 hasConceptScore W2170938943C120665830 @default.
• W2170938943 hasConceptScore W2170938943C121332964 @default.
• W2170938943 hasConceptScore W2170938943C14999030 @default.
• W2170938943 hasConceptScore W2170938943C154945302 @default.
• W2170938943 hasConceptScore W2170938943C177264268 @default.
• W2170938943 hasConceptScore W2170938943C192209626 @default.
• W2170938943 hasConceptScore W2170938943C195324797 @default.
• W2170938943 hasConceptScore W2170938943C199360897 @default.
• W2170938943 hasConceptScore W2170938943C204321447 @default.
• W2170938943 hasConceptScore W2170938943C2776224158 @default.
• W2170938943 hasConceptScore W2170938943C28490314 @default.
• W2170938943 hasConceptScore W2170938943C41008148 @default.
• W2170938943 hasConceptScore W2170938943C53893814 @default.
• W2170938943 hasConceptScore W2170938943C542774811 @default.
• W2170938943 hasConceptScore W2170938943C60048249 @default.
• W2170938943 hasLocation W21709389431 @default.
• W2170938943 hasOpenAccess W2170938943 @default.
• W2170938943 hasPrimaryLocation W21709389431 @default.
• W2170938943 hasRelatedWork W1479882947 @default.
• W2170938943 hasRelatedWork W1525797625 @default.
• W2170938943 hasRelatedWork W1972873790 @default.
• W2170938943 hasRelatedWork W1989768143 @default.
• W2170938943 hasRelatedWork W2034090682 @default.
• W2170938943 hasRelatedWork W2047708730 @default.
• W2170938943 hasRelatedWork W2069059998 @default.
• W2170938943 hasRelatedWork W2124879043 @default.
• W2170938943 hasRelatedWork W2130416582 @default.
• W2170938943 hasRelatedWork W2136417990 @default.
• W2170938943 hasRelatedWork W2145699207 @default.
• W2170938943 hasRelatedWork W2319236875 @default.
• W2170938943 hasRelatedWork W2532440341 @default.
• W2170938943 hasRelatedWork W2578832979 @default.
• W2170938943 hasRelatedWork W2742039928 @default.
• W2170938943 hasRelatedWork W2785466268 @default.
• W2170938943 hasRelatedWork W2810383139 @default.
• W2170938943 hasRelatedWork W2952712483 @default.
• W2170938943 hasRelatedWork W3100400819 @default.
• W2170938943 hasRelatedWork W111538421 @default.
• W2170938943 isParatext "false" @default.
• W2170938943 isRetracted "false" @default.
• W2170938943 magId "2170938943" @default.
• W2170938943 workType "article" @default.