Dissertations / Theses on the topic 'Govorin'
Create a spot-on reference in APA, MLA, Chicago, Harvard, and other styles
Consult the top 19 dissertations / theses for your research on the topic 'Govorin.'
Next to every source in the list of references, there is an 'Add to bibliography' button. Press on it, and we will generate automatically the bibliographic reference to the chosen work in the citation style you need: APA, MLA, Harvard, Chicago, Vancouver, etc.
You can also download the full text of the academic publication as pdf and read online its abstract whenever available in the metadata.
Browse dissertations / theses on a wide variety of disciplines and organise your bibliography correctly.
Vanja, Barišić-Joković. "Kratki govorni izrazi usmenog porekla u skupštinskog govoru – zastupljenost, značenje, funkcija." Phd thesis, Univerzitet u Novom Sadu, Doktorske disertacije iz interdisciplinarne odnosno multidisciplinarne oblasti na Univerzitetu u Novom Sadu, 2016. https://www.cris.uns.ac.rs/record.jsf?recordId=100255&source=NDLTD&language=en.
Full textThis doctoral dissertation is dedicated to the research of spoken expressions of the verbal origin in the speech of the National Assembly of the Republic of Serbia, founded on the base which comprises of 19 years of the Assembly sessions, 1997 to 2015.Thematic and theoretical – methodological area of the research, as well as the basic unit of the analysis: spoken expression of the verbal origin have been derived from the dissertation title SPOKEN EXPRESSIONS OF THE VERBAL ORIGIN IN THE ASSEMBLY SPEECH – OCCURENCE, MEANING, FUNCTION. Considering that relatively few national scientific papers are dedicated to spoken expressions of the verbal origin in the contemporary / political speech, the aim of this dissertation was to create a base of such phrases and to analyze them in the context as well as to study their occurrence, meaning and function.A part of inspiration for this research was an innovative work of Gary Alan Fine, which offered to folkloristics sociological and socio-psychological methods which interpreted folklore stories as segments of a social system; to sociology, it offered ethnography of small groups, a study of microstructures and miniature social phenomena which can assist in generating the answers in macrosociology. In the light of his theory - conduit theory, there was neither more prominent place than the National Assembly, nor more prominent social group than members of the parliament, for this research. The issue which this dissertation is aiming to address is how the social and political elite, in the most significant institution deals with the segment of national tradition. This issue according to Fine's theory can also raise some more significant questions related to the whole society.
Siniša, Suzić. "Parametarska sinteza ekspresivnog govora." Phd thesis, Univerzitet u Novom Sadu, Fakultet tehničkih nauka u Novom Sadu, 2019. https://www.cris.uns.ac.rs/record.jsf?recordId=110631&source=NDLTD&language=en.
Full textIn this thesis methods for expressive speech synthesis using parametricapproaches are presented. It is shown that better results are achived withusage of deep neural networks compared to synthesis based on hiddenMarkov models. Three new methods for synthesis of expresive speech usingdeep neural networks are presented: style codes, model re-training andshared hidden layer architecture. It is shown that best results are achived byusing style code method. The new method for style transplantation based onshared hidden layer architecture is also proposed. It is shown that thismethod outperforms referent method from literature.
Targhetta, Francesco. "Corrado Govoni 1903-1907." Doctoral thesis, Università degli studi di Padova, 2008. http://hdl.handle.net/11577/3425539.
Full textGovori, Bukurije [Verfasser]. "Studien zur Synthese der Naturstoffe Sorangicin Z und Haprolid / Bukurije Govori." Hannover : Technische Informationsbibliothek und Universitätsbibliothek Hannover (TIB), 2015. http://d-nb.info/1072062550/34.
Full textStevan, Ostrogonac. "Modeli srpskog jezika i njihova primena u govornim i jezičkim tehnologijama." Phd thesis, Univerzitet u Novom Sadu, Fakultet tehničkih nauka u Novom Sadu, 2018. https://www.cris.uns.ac.rs/record.jsf?recordId=107812&source=NDLTD&language=en.
Full textA statistical language model, in theory, represents a probability distribution over sequences of words of a language. In practice, it is a tool for estimating probabilities of word sequences of interest. Mathematical basis related to language models is mostly language independent. However, the quality of trained models depends not only on training algorithms, but on the amount and quality of available training data as well. For languages with complex morphology, such as Serbian, textual corpora for training language models need to be significantly larger than the corpora needed for training language models for languages with relatively simple morphology, such as English. This research represents the entire process of developing language models for Serbian, starting with collecting and preprocessing of textual contents, extending to adaptation of algorithms and development of methods for addressing the problem of insufficient training data, and finally to adaptation and application of the models in different technologies, such as text-to-speech synthesis, automatic speech recognition, automatic detection and correction of grammar and semantic errors in texts, and determining basics for the application of the models in automatic document classification and other tasks. The core of the development of language models for Serbian is defining morphologic classes of words, based on the information contained within the morphologic dictionary of Serbian, which was one of the results of a previous research.
Mila, Veselinović. "Говорни капацитет особа са израштајима на гласницама." Phd thesis, Univerzitet u Novom Sadu, Medicinski fakultet u Novom Sadu, 2017. http://www.cris.uns.ac.rs/record.jsf?recordId=104440&source=NDLTD&language=en.
Full textGovor predstavlja vrlo složenu ljudsku delatnost. Za njegovu uspešnu realizaciju potrebno je uskladiti pravilnu artikulaciju, ispravan akustički signal, nesmetanu percepciju i odgovarajuću kognitivnu spoznaju. Kao najznačajniji modalitet realizacije jezika u procesu komunikacije u velikoj meri zavisi od kvalitetaglasa koji je njegov neodvojivi deo. Patološke karakteristike glasa i govora osoba sa benignim i malignim izraštajima na glasnicama percipiraju se kao različite varijacije visine, intenziteta i kvaliteta govornog glasa. Ciljevi istraživanja: Utvrditi govorni kapacitet osoba sa malignim, benignim i pseudotumorima, na glasnicama pre i posle operativnog lečenja, razliku između njihovog govornog kapaciteta i utvrditi kako operativno lečenje utiče na razumljivost i vremenske odrednice govora. Istraživanje je sprovedeno tokom 2016. i 2017. godinena Klinici za bolesti uva, grla i nosa Kliničkog centra Vojvodine. Uzorak je činilo 67 ispitanika podeljenih u 2 grupe prema vrsti izraštaja na one koji su imali benigne i pseudotumore i ispitanike sa malignim tumorima na glasnicama starosti od 23 do 74 godine (AS 55,43 godine; SD 11,95 godina). Metode primenjene u istraživanju: Opšti upitnik, Testovi za isključivanje iz uzorka (Mini-Mental test - Mini mental State Examination, Upitnik o anksioznosti-GAD-7 Anxiety, Trijažni artikulacioni test), Indeks govornog hendikepa -Speech Handicap Index,Akustička analiza glasa, Maksimalno fonacijsko vreme vokala /a/, Analiza vremenske organizacije govora, Razumljivost govora, Analiza melodije rečenice. Prema našim rezultatima, pre operacije u obe grupe ispitanika govorni kapacitet je bio gotovo ujednačen sa malom razlikom jer su ipak nešto bolji govorni kapacitet imale osobe sa malignim izraštajima na glasnicama ali bez statističke značajnosti u poređenim grupama. Posle operativnog lečenja benignih i pseudotumora na glasnicama došlo je do statistički značajnog poboljšanja govornog kapaciteta (p=0,005). Posle operacije postojalaje statistički značajna razlika između govornog kapaciteta ispitivanih grupa, pri čemu su bolji govorni kapacitet imali ispitanici grupe I – sa benignim i pseudotumorima (t=-3,807, p<0,001). Razumljivost govora je kod ispitanika sa malignim tumorima glasnice bila statistički značajno povezana sa vrednostima govornog kapaciteta pre operacije (p=0,008). U pogledu vremenske organizacije govora, kod ispitanika sa malignim tumorima na glasnicama posle operacije došlo je do statistički značajnog pogoršanja (p=0,025). U istraživanju nije dobijen izolovani efekat vremena ili vrste izraštaja na govorni kapacitet, ali postoji njihov udruženi efekat (f = 10,079, p = 0,002). Preporuka je da se svi pacijenti posle operativnog lečenja izraštaja na glasnicama uključe u postupak rehabilitacije glasa i govora.
Speech represents a very complicated human ability. For its successful realization, it’s necessary to coordinate proper articulation, correct acoustical signal, unobstructed perception and adequate cognitive recognition. The most significant modality of language realization in the process of communication greatly depends on vocal quality, which is its inseparable element. Pathological vocal and speech characteristics in people with benign and malign growths on the vocal cords are perceived as different variations of pitch, intensity and quality of the speaking voice. Aims of the research: Determining the vocal capacity of people with malign, benign tumors and pseudo-tumors on the vocal cords, before and after surgical treatment, determining the difference between their vocal capacity and how surgical treatment affects comprehensibility and time determinants of speech. The research was conducted during 2016 and 2017 on the Otorhinolaryngology clinic of Vojvodina Clinical Centre. The sample consisted of 67 participants divided into 2 groups based on the type of growth present, into participants with benign and pseudo tumors and participants with malign tumors of the vocal chords, aged 23 – 74 (M=55,43; SD=11,95). Methods used in the research: General Questionnaire, Tests for exclusion from the sample (Mini mental State Examination), Anxiety Questionnaire – GAD-7 Anxiety, Triage Articulatory Test, Speech Handicap Index, Acoustic Vocal Analysis, Maximum phonation time vowel /a/, Time organization of speech analysis, Speech comprehensibility, Sentence melody analysis. According to the results of the research, before surgical treatment, the vocal capacity of both participant groups was nearly equable with a small difference of a somewhat higher vocal capacity in participants with malign growths on the vocal cords, but with no statistical significance between the compared groups. After surgical treatment of benign and pseudo tumors on the vocal cords, a statistically significant improvement of vocal capacity was observed (p=0,005). Post surgically, a statistically significant difference between the vocal capacity of tested groups was observed, with a higher vocal capacity present in the first group of participants with benign and pseudo tumors (t=-3,807, p<0,001). Speech comprehensibility in participants with malign tumors of the vocal cords was significantly correlated with the values of speech capacity prior to the surgical treatment (p=0,008). In regards to time organization of speech, there was a statistically significant deterioration in patients with malign vocal cord tumors (p=0,025). The research showed no isolated effect of time or type of growth on speech capacity, but a combined effect was observed (f=10,079, p=0,002). The recommendation is that all patients are included in the procedure of speech and vocal rehabilitation after surgical treatment.
Sandra, Sovilj-Nikić. "Razvoj matematičkog modela trajanja glasova u automatskoj sintezi govora na srpskom jeziku." Phd thesis, Univerzitet u Novom Sadu, Fakultet tehničkih nauka u Novom Sadu, 2014. https://www.cris.uns.ac.rs/record.jsf?recordId=85851&source=NDLTD&language=en.
Full textIn this dissertation several different phone duration models of the Serbainlanguage using appropriate machine learning algorithms were developed.The objective evaluation of the models obtained and their mutual comparisonbased on quantitative measures such as RMSE (root-mean-squared error),MAE (mean absolute error) and CC (correlation coefficient) were performed.The comparison of the models developed for the Serbian language with theperformances of the models developed for other languages is also carriedout. It was observed that the performances of the models developed in thisdissertation are comparable or even outperform the performances of themodels that have been developed for other languages.
Bianchi, Matteo <1987>. "La poesia ferrarese dopo Corrado Govoni, dai crepuscoli sul Po agli influssi emiliani." Master's Degree Thesis, Università Ca' Foscari Venezia, 2015. http://hdl.handle.net/10579/6962.
Full textNikša, Jakovljević. "Primena retke reprezentacije na modelima Gausovih mešavina koji se koriste za automatsko prepoznavanje govora." Phd thesis, Univerzitet u Novom Sadu, Fakultet tehničkih nauka u Novom Sadu, 2014. http://dx.doi.org/10.2298/NS20131218JAKOVLJEVIC.
Full textThis thesis proposes a model which approximates full covariance matrices inGaussian mixture models with a reduced number of parameters andcomputations required for likelihood evaluations. In the proposed modelinverse covariance (precision) matrices are approximated using sparselyrepresented eigenvectors. A maximum likelihood algorithm for parameterestimation and its practical implementation are presented. Experimentalresults on a speech recognition task show that while keeping the word errorrate close to the one obtained by GMMs with full covariance matrices, theproposed model can reduce the number of parameters by 45%.
Edvin, Pakoci. "Uticaj morfoloških obeležja na modelovanje jezika primenom neuronskih mreža u sistemima za prepoznavanje govora." Phd thesis, Univerzitet u Novom Sadu, Fakultet tehničkih nauka u Novom Sadu, 2019. https://www.cris.uns.ac.rs/record.jsf?recordId=111244&source=NDLTD&language=en.
Full textAutomatic speech recognition is a technology that allows computers toconvert spoken words into text. It can be applied in various areas whichinvolve communication between humans and machines. This thesis primarilydeals with one of two main components of speech recognition systems - thelanguage model, that specifies the vocabulary of the system, as well as therules by which individual words can be linked into sentences. The Serbianlanguage belongs to a group of highly inflective and morphologically richlanguages, which means that it uses a number of different word endings toexpress the desired grammatical, syntactic, or semantic function of the givenword. Such behavior often leads to a significant number of errors in speechrecognition systems where due to good acoustic matching the recognizercorrectly guesses the basic form of the word, but an error occurs in the wordending. This word ending may indicate a different morphological category, forexample, word case, grammatical gender, or grammatical number. Thethesis presents a new language modeling tool which, along with the wordidentity, can also model additional lexical and morphological features of theword, thus testing the hypothesis that this additional information can helpovercome a significant number of recognition errors that result from the highinflectivity of the Serbian language.
Radmila, Radonjić. "Engleski modali u pravnim dokumentima Evropske unije i njihovi srpski ekvivalenti." Phd thesis, Univerzitet u Novom Sadu, Filozofski fakultet u Novom Sadu, 2018. https://www.cris.uns.ac.rs/record.jsf?recordId=107081&source=NDLTD&language=en.
Full textOva disertacija predstavlja ispitivanje modalnih glagola i kategorije modalnosti u okviru pravnog funkcionalnog stila i teorije prevođenja. Naime, engleski modali proučavaju se na korpusu pravnih dokumenata Evropske unije sa akcentom na prevodne ekvivalente u srpskom jeziku. Modali se proučavaju prvenstveno kao eksponentni deontičke modalnosti u okviru koje se pripisuju značenja obaveze, naredbe, zabrane, dozvole, zahtjeva i slično, što je u skladu sa semantikom pravnog diksursa koji je podložan modalizaciji.
Modali se proučavaju na korpusu dokumenata Evropske unije, s jedne strane zbog činjenice da su pravne zakonitosti institucionalizovane u okviru Evropske unije nešto drugačije normirane od uobičajenog pravnog registra anglosaksonskog pravnog sistema, a s druge strane zbog aktuelne situacije u pogledu evrointegracija i potrebe usklađivanja domaćeg zakonodavstva sa zakonodavstvom Evropske unije, a samim tim i prevođenja velikog broja dokumenata sa engleskog na srpski jezik i obrnuto. Interesovanje za proučavanje modalnih glagola u okviru pravnog funkcionalnog stila javilo se zbog kompleksnosti upotrebe modalnih glagola u savremenom engleskom jeziku, ali i zbog niza pravila koja ih prate u okviru jezičke politike Evropske unije i niza problema sa kojima se prevodioci suočavaju prilikom prevođenja pravnih dokumenata. Pravni tekstovi podliježu značajnim ograničenjima zbog specifične terminologije i odgovarajućih sintaksičkih struktura, a budući da prevodioci najčešće nijesu pravnici neophodno je istaći sve specifičnosti prevođenja pravnih propisa sa akcentom na modalnim glagolima.
Modalnost je semantičko-gramatička kategorija koja se izražava na različite načine u različitim jezicima, a budući da je riječ o kontrastivnom istraživanju, cilj rada je i da se kontrastivnom metodom utvrdi šta engleski i srpski jezik imaju zajedničko kod upotrebe modalnih glagola u pravnom registru, kao i koje se razlike javljaju među njima. To znači da primarni cilj istraživanja predstavlja popisivanje svih modalnih glagola u korpusu, potom njihovu analizu sa aspekta semantike, a zatim utvrđivanje na koji način su prevedeni na srpski jezik, koje se zakonitosti javljaju u procesu prevođenja, koje tekstualne zakonitosti i ograničenja postoje s obzirom na specifičnosti kontrastiranih jezika. Budući da se u fokusu rada nalazi jedna semantički i sintaksički veoma složena kategorija u radu je primijenjen lingvističko- semantički pristup. Teorijski okvir i interpretacija značenja zasnivaju se na
tradicionalnoj i savremenoj literaturi o modalnosti i modalnim glagolima u ispitivanim jezicima.
Disertacija se posebno bavi sljedećim pitanjima: koliko različiti žanrovi, odnosno tipovi diskursa utiču na modalnost; koja vrsta modalnosti se najčešće javlja u pravnim tekstovima; koje su sličnosti i razlike između engleskog i srpskog jezika s obzirom na ovaj koncept; da li bi bilo moguće sačiniti skalu poklapanja i nepoklapanja između engleskog i srpskog na kontrastivnoj osnovi sa posebnim osvrtom na prevođenje modalnih glagola u pravnim tekstovima; kako se tipovi modalnosti prevode s jednog na drugi jezik kad su u pitanju pravni tekstovi; u kom jeziku se više upotrebljavaju modalni glagoli u diskursu prava, engleskom ili srpskom; koja morfosintaksička sredstva se koriste više ili manje često za izražavanje modalnosti; kako se svaki konkretni modalni glagol prevodi, da li se najčešće zamjenjuje najbližim ekvivalentom u drugom jeziku. Osnovni metodološki pristup primijenjen u istraživanju je deskriptivni, s tim što je analiza i kvalitativnog i kvantitativnog karaktera.
Dragiša, Mišković. "Контекстно зависно препознавање говора у интеракцији између човека и машине." Phd thesis, Univerzitet u Novom Sadu, Fakultet tehničkih nauka u Novom Sadu, 2017. https://www.cris.uns.ac.rs/record.jsf?recordId=104147&source=NDLTD&language=en.
Full textPored velikog značaja kontekstualnih informacija pri razumevanjugovora, njihova obrada i upotreba u savremenim sistemima zaautomatsko prepoznavanje govora je veoma ograničena, što znatnonarušava performanse prepoznavanja u realnim uslovima upotrebe.Stoga, ukoliko želimo da se karakteristike ovih sistema približeljudskim, neophodno je uključiti kontekst u adekvatnom obimu.U ovoj tezi je predstavljen novi metodološki pristup kontekstnozavisnom prepoznavanju govora u interakciji između čoveka i mašine.Na metodološkom nivou, ovaj pristup je hibridan, jer integrišestatističke i simboličke metode, i kognitivno inspirisan, jer uzima uobzir uvide u rezulatate is traživanja iz oblasti neurokognitivnihnauka. Osnovni princip je da se ocenjivanje hipoteza sistema zaprepoznavanje vrši na osnovu njihove kontekstualne usklađenosti,informacionog sadržaja i semantičke ispravnosti.Pristup je ilustrovan prototipskim implementacijama za konkretnedomene interakcije.
Although the importance of contextual information in speech recognition hasbeen acknowledged for a long time now, it remained clearly underutilizedeven in state-of-the-art speech recognition systems. This thesis introduces anovel, methodologically hybrid approach to the research question of contextdependentspeech recognition in human-machine interaction. To the extentthat it is hybrid, the approach integrates aspects of both statistical andrepresentational paradigms. The aim of this thesis is to extend the standardstatistical pattern matching approach with a cognitively-inspired andanalytically tractable model with explanatory power. This methodologicalextension allows for accounting for contextual information which is otherwiseunavailable in speech recognition systems, and using it to improve postprocessingof recognition hypotheses. The thesis introduces an algorithm forevaluation of recognition hypotheses, illustrates it for concrete interactiondomains, and discusses its implementation within two prototypeconversational agents.
Zoran, Simić. "Синтакса падежа у романијским говорима Старог Влаха: метафоризација простора." Phd thesis, Univerzitet u Novom Sadu, Filozofski fakultet u Novom Sadu, 2016. http://www.cris.uns.ac.rs/record.jsf?recordId=100379&source=NDLTD&language=en.
Full textThe thesis analyses the function ofdeterminative case systems in the Romanijaspeech of the Stari Vlah region. It deals withsyntactic and semantic characteristics of caseconstructions which act as syntactic formalisersof the semantic category of spatiality as well asthose which are syntactic formalisers of moreabstract semantic categories, such as thefollowing: temporality, qualificability(including instrumentality, qualificability in anarrower sense, concomitancy, the criterion /basis of the qualificative type, comparabilityand quantifiability), intentionality, causality,conditionality, concessionality andpossessiveness. A special attention is devoted tothe case constructions whose primary meaningis that of spatiality, as their source domain,which then by the mechanism ofmetaphorisation of space find their place in thesystem of formalisers of the aforementionedabstract semantic categories as their targetdomains.
Branko, Lučić. "Развој мултимодалног система образовне игре за слабовиду и слепу децу." Phd thesis, Univerzitet u Novom Sadu, Fakultet tehničkih nauka u Novom Sadu, 2016. https://www.cris.uns.ac.rs/record.jsf?recordId=101075&source=NDLTD&language=en.
Full textU disertaciji su prikazani razvoj multimodalnogsistema za obrazovnu igru Lugram i istraživanjavezana za njegovu primenu u oblasti redovnog ispecijalizovanog osnovnoškolskog obrazovanja.Izvršeno je ispitivanje uticaja Lugrama napovišenje efekata učenja geometrije u redovnojrazrednoj nastavi, ispitivanje da li njegovuprilagođenu multimodalnu verziju mogu da koristeslabovida i slepa deca i da li se ona možekoristiti za njihovu obuku za samostalnu upotreburačunarske audio verzije Lugrama.Utvrđeno je da Lugram ima uticaja na porast uspehau učenju geometrije i da se prilagođena audio-taktilna verzija može koristiti u svrhu obukeslabovide i slepe dece za samostalno korišćenjeaudio verzije igre.Rezultati istraživanja su pokazali opravdanostpristupa razvoju Lugrama kao multimodalnogsistema za igru i usmerili njegov dalji razvoj kaasistivnom multimodalnom sistemu koji posedujesposobnost govorne interakcije sa korisnikom imogućnost prilagođenja različitim kategorijamakorisnika.
The dissertation presents the development of the multimodal system for the educational game Lugram and the research related to its application in the regular and specialized primary education. The examination of the effect which Lugram has on the improvement in learning geometry in the primary education is done, as well as the testing weather its adapted multimodal version can be used for the training of partially sighted and blind children for the independent use of the computer audio version of Lugram. It is established that Lugram has an influence on the improvement in learning geometry and that adapted audio-tactile version can be used for the training purposes for partially sighted and blind children. Results of the research have justified the approach in the development of Lugram as a multimodal system for the game. Moreover, they focus further development toward assistive multimodal system which has ability of the voice interaction with the user and also ability of adaptation to the different kind of users.
Branislav, Popović. "Хијерархијско кластеровање модела Гаусових смеша у апликацијама за континуално препознавање говора." Phd thesis, Univerzitet u Novom Sadu, Fakultet tehničkih nauka u Novom Sadu, 2012. http://dx.doi.org/10.2298/NS20120717POPOVIC.
Full textU okviru doktorske disertacijepredstavljen je novi algoritamhijerarhijskog klasterovanja modelaGausovih smeša, zasnovan na operacijipodele i spajanja. Algoritam težipoboljšanju lokalno optimalnogrešenja određenog inicijalnomkonstelacijom. Inicijalizuje selokalno optimalnim parametrima,dobijenim korišćenjem referentnogpristupa sličnog k‐means‐u i teži kapribližavanju globalnom optimumufunkcije cilja, iterativnomprimenom operacija podele i spajanjanad klasterima Gausovih komponenti,određenim primenom referentnogalgoritma.Dodatno poboljšanje algoritmaosvareno je uvođenjem principaselekcije modela u svrhu određivanjanajpovoljnijeg odnosa tačnosti iračunske složenosti, u zadatku selekcije gausijana unutar realnogsistema za prepoznavanje. Predloženimetod testiran je kako nad veštačkigenerisanim podacima, tako i uokviru algoritma selekcije gausijana,na primeru sistema za kontinualnoprepoznavanje govora. U oba slučajazabeleženo je poboljšanje rezultata uodnosu na rezultate ostvareneprimenom referentnog algoritma.Poboljšanja algoritma selekcijegausijana izborom optimalnog skupasistemskih parametara takođe surazmotrena.
The dissertation presents a novel splitand‐merge algorithm for hierarchicalclustering of Gaussian mixture models.The algorithm tends to improve on thelocal optimal solution determined by theinitial constellation. It is initialized bylocal optimal parameters obtained byusing a baseline approach similar to kmeans,and it tends to approach moreclosely to the global optimum of thetarget clustering function, by iterativelysplitting and merging the clusters ofGaussian components obtained as theoutput of the baseline algorithm.The algorithm is further improved byintroducing model selection in order toobtain the best possible trade‐offbetween recognition accuracy andcomputational load in a Gaussianselection task applied within an actualrecognition system. The proposedmethod is tested both on artificial dataand in the framework of Gaussianselection performed within a real continuous speech recognition system. Inboth cases an improvement over thebaseline method has been observed.Additional improvements of Gaussianselection algorithm by using the optimalset of system parameters are alsodiscussed.
Boško, Božilović. "Биометријско обележје за препознавање говорника: дводимензионална информациона ентропија говорног сигнала." Phd thesis, Univerzitet u Novom Sadu, Fakultet tehničkih nauka u Novom Sadu, 2016. http://www.cris.uns.ac.rs/record.jsf?recordId=101369&source=NDLTD&language=en.
Full textMotiv za istraživanje je unapređenje procesa automatskog prepoznavanja govornika bez obzira na sadržaj izgovorenog teksta.Cilj ove doktorske disertacije je definisanje novog biometrijskog obeležja za prepoznavanje govornika nezavisno od izgovorenog teksta − dvodimenzionalne informacione entropije govornog signala.Definisanje novog obeležja se vrši isključivo u vremenskom domenu, pa je računarska složenost algoritma za njegovo izdvajanje znatno manja u odnosu na obeležja koja se izdvajaju u frekvencijskom domenu. Ocena performansi dvodimenzionalne informacione entropije je urađena nad reprezentativnim skupom slučajno odabranih govornika. Pokazano je da predloženo obeležje ima malu varijabilnost unutar govornog signala jednog govornika, a veliku varijabilnost između govornih signala različitih govornika.
Тhe motivation for the research is the improvement of the automatic speaker recognition process regardless of the content of spoken text.The objective of this dissertation is to define a new biometric text-independent speaker recognition feature − the two-dimensional informational entropy of speech signal.Definition of the new feature is performed in time domain exclusively, so the computing complexity of the algorithm for feature extraction is significantly lower in comparison to feature extraction in spectral domain. Performance analysis of two-dimensional information entropy is performed on the representative set of randomly chosen speakers. It has been shown that new feature has small within-speaker variability and significant between-speaker variability.
Nataša, Cvijanović. "Предшколски програм у Републици Српској као чинилац учења и развоја дјеце." Phd thesis, Univerzitet u Novom Sadu, Filozofski fakultet u Novom Sadu, 2016. http://www.cris.uns.ac.rs/record.jsf?recordId=100349&source=NDLTD&language=en.
Full textRad je nastao kao rezultat teorijskog i em-pirijskog istraživanja efekta pohađanjapredškolskog programa u Republici Srp-skoj na učenje i razvoj djece predškolskoguzrasta.U teorijskom dijelu razmatraju se pitanja opredškolskom programu kao osnovi insti-tucionalnog predškolskog vaspitanja i ob-razovanja, o pojmovnom razgraničenju pojmo-va program i kurikulum, (o određenju) tradi-cionalnog naspram savremenog predškol-skog programa/kurikuluma, sistemu pred-školskog vaspitanja i obrazovanja u Repu-blici Srpskoj, kao i o odnosu programa ivaspitne prakse. Takođe, posmatrana su ipitanja koja se odnose na rano učenje, počevod toga šta je rano učenje i kako ga defini-sati, preko shvatanja o ranom učenju krozistoriju, karakteristika ranog učenja, odno-sa ranog učenja i vaspitanja, te društveneparadigme ranog vaspitanja, kao i pitanjakoja se odnose na vezu dječje aktivnosti,igre i ranog učenja, pa sve do značaja i kori-sti od ranog učenja za pojedinca i društvo.U dijelu rada koji se odnosi na empirijskipristup problemu istraživanja prikazanisu rezultati istraživanja i nalazi dobijeniispitivanjem međuzavisnosti pohađanja pro-grama predškolskog vaspitanja i obrazova-nja i razvoja predškolske djece posmatranogkroz fizički, socio-emocionalni, intelek-tualni razvoj i razvoj govora, komunikacijei stvaralaštva.Istraživanje se temelji na hipotezi da iz-među pohađanja predškolskog programa i ra-nog učenja postoji značajna i izražena među-zavisnost. Drugim riječima, istraživanjemsmo htjeli utvrditi da li pohađanje pred-školskog programa determiniše rano uče-nje kod djece. Postavljeni ciljevi i zadaci istraživanjarealizovani su, kako smo već rekli, kroz te-orijsko istraživanje, ali i kroz empirijskoneeksperimentalno istraživanje, primje-nom instrumenata koji su konstruisani zapotrebe ovog rada. Prvi instrument odnosise na utvrđivanje ispoljenosti razvojnihaspekata djece, dok se drugi instrument od-nosi na prikupljanje podataka o odnosu po-rodičnih prilika i razvojnih aspekata dje-ce. Prvi instrument bio je namijenjen uči-teljima, a drugi roditeljima.Uzorak je obuhvatio 1.439 djece koja su kre-nula u prvi razred (šest godina), i to: djecukoja su pohađala neku predškolsku ustanovui djecu koja nisu pohađala predškolskuustanovu niti su bila uključena u neki ob-lik predškolskog vaspitanja i obrazovanja.Isto tako, u istraživanju su učestvovali iroditelji ispitivane djece, od kojih smo do-bili podatke o socijalnom statusu porodicedjeteta i njihovoj posvećenosti roditelj-stvu.Rezultati koje smo dobili dijelom su potvr-dili hipoteze koje smo postavili, a timenas i usmjerili na dodatna razmišljanja oefektima pohađanja predškolskog programana sve aspekte razvoja, ponajviše na socio-emocionalni razvoj, jer je ovaj aspekt razvo-ja visoko razvijen kod obje grupe djece: kojasu pohađala predškolski program i koja ganisu pohađala.
In search of the new answers to already knownquestions concerned with discovering the causesof different development of certain abilities inchildren, this study ideas are focused on the interdependenceof stimulating learning environmentand early learning for children. This studyis a result of theoretical and empirical researchof the effect of attending preschool programmein Republic of Srpska on learning and developmentof children of preschool age.The theoretical part investigates questions onpreschool programme as the foundation of institutionalpreschool education, concept delimitationof programme and curriculum, (definitionof) traditional versus contemporary preschoolprogramme/curriculum, system of preschooleducation in Republic of Srpska, as well as questionsabout relationship between children’s activities,play and early learning. Furthermore,this paper investigates questions related to earlylearning, starting from the notion of early learningand how to define it, the understanding ofearly learning throughout history, early learningcharacteristics, the relationship of early learningand education as well as social paradigm ofearly education, questions related to the relationshipbetween children’s activities, play andearly learning, to the importance and benefits ofearly learning for an individual and a society.A part of the study related to the empirical approachto the research topic presents the researchresults and findings obtained by examinationof the interdependence of attending thepreschool education and development ofpreschool children, observed through physicaland socio-emotional development and developmentof speech, communication and creativity.This study is based on the hypothesis that thereis a significant and profound interdependencebetween attending a preschool programme and13early learning. In other words, our intention wasto establish whether attending a preschool programmedetermines early learning for children.As mentioned before, the established researchgoals and objectives have been accomplishedthrough theoretical research but also throughnon-experimental empirical research, using instrumentsconstructed for this study. First instrumentrefers to the expression of developmentalaspects of children while the second instrumentrefers to data collection on relationshipbetween family context and developmentalaspects of children. First instrument was intendedfor teachers and second one for parents.The sample included 1 439 children who startedfirst grade (6 years) as follows: children who attendeda preschool institution and children whodid not attend preschool institutions nor wereinvolved in any form of preschool education.Furthermore, the research included children’sparents who provided the information on child’sfamily social status and their own devotion toparenthood.The results obtained have partly confirmed thehypothesis set and therefore directed us to furtherreflection on the effects of attending a preschoolprogram on all aspects of development,especially socio-emotional development sincethis aspect is highly developed with both groupsof children: those who attended a preschoolprogramme and those who did not.
Ružica, Mirilov. "Кулинарска терминологија Војводине." Phd thesis, Univerzitet u Novom Sadu, Filozofski fakultet u Novom Sadu, 2016. http://www.cris.uns.ac.rs/record.jsf?recordId=101013&source=NDLTD&language=en.
Full textKulinarska leksika uopšte, kao ikulinarska leksika Vojvodine nije do sadadetaljnije sa leksičko semantičkog itvorbenog aspekta obrađivana (osim upojedinim prikazima naših dijelektologa ileksikografa), ali je svakako ona (ne upotpunosti) deo srpskih važnijihdijalekatskih rečnika (Od Vukova doRečnika srpskih govora Vojvodine) irečnika opšteg tipa.U istraživačkom radu je primenjena u nasodomaćena teorija semantičkih polja ruskogetnolingviste Nikite Tolstoja. U leksičkosemantičkoj i tvorbenoj analizi kao i uleksikografskoj obradi primenjen jedeskriptivni, lingvogeografski ikomparativni metod.Obrađeno je šesnaest semantičkih polja:Namirnice, Obroci, Pripremanje hrane,Zimnica, Mleko , mlečni proizvodi i jelaod mleka, Supe i čorbe, Jela od testa,Jela od jaja, Jela od mesa, Mesneprerađevine, Vrste mesa i delovi, Umaci,Jela od povrća, Kolači, slatkiši, Pića,Jela prilikom raznih svetkovina.Materijal je i leksikografski uobličen, apotom kartografisan.
Culinary lexicon as such, as well as theculinary lexicon of Vojvodina Region has so farnot been systematically examined in the lexicosemanticand word-formation aspect (except forsome reviews of our dialectologists and lexicographers). However, it is (yet notcompletely) a part of some major Serbiandialect dictionaries (from Rječnik, a dictionaryby Vuk Stefanović Karadžić to Dictionary ofSerbian Vojvodina Speech) as well as therange of general dictionaries.The research implied the semantic fieldtheory developed by Russian ethno-linguistNikita Tolstoy, which is commonly appliedapproach in our region. Lexico-semantic andword-formation analysis as well aslexicographic processing were performedapplying descriptive and linguo-geographicmethods.Sixteen semantic fields were analyzed:Foodstuffs; Meals; Food Preparation; WinterStores; Milk; Dairy Products and Dishes; Soupsand Stocks; Pasta and Savory Dishes; EggDishes; Meat Dishes; Meat Products; MeatTypes and Cuts; Sauces; Vegetable Dishes;Cakes, Cookies and Sweets; Drinks andBeverages; Festive Foods. The material waslexicographically edited and cartographed.
Gvozdic, Davor. ""Olá senhor presidente e obrigado por ter vindo!" : os atos de saudação e de agradecimento no discurso diplomático : estudo contrastivo português - croata." Master's thesis, 2012. http://hdl.handle.net/10400.2/2597.
Full textA importância de reconhecer a intenção comunicativa dos atores numa interação verbal revela-se diariamente na prática de tradução-interpretação, especialmente quando os tradutores-intérpretes exercem a sua atividade profissional numa situação enunciativa delicada como é a que se associa ao discurso diplomático. O presente estudo, no âmbito da Pragmática Linguística, resulta de uma investigação, que, numa perspetiva contrastiva, observa a realização dos atos de saudação e dos atos de agradecimento, no contexto do discurso diplomático, em duas línguas – português europeu e croata. O objetivo é tentar compreender os processos linguísticos subjacentes à realização dos referidos atos ilocutórios, estabelecendo um contraste entre os resultados obtidos em português e em croata para, através desta análise contrastiva, determinar algumas características gerais do discurso diplomático. Para tal, parte-se da observação detalhada do objeto de estudo central, um conjunto de dois corpora de textos do âmbito específico do discurso diplomático: um subcorpus em português e outro em croata. Os resultados obtidos demonstram uma maior tendência para a aproximação do que para o distanciamento das duas línguas quanto ao modo de realização dos atos de saudação e dos atos de agradecimento no contexto analisado. Assim, é possível definir os princípios gerais que, no contexto do discurso diplomático, regem a produção verbal dos interactantes. Como resultado prático, o presente estudo apresenta uma proposta de esquematização das formas linguísticas através das quais são realizados os atos de saudação e os atos de agradecimento, no referido contexto.
The work of a translator-interpreter requires that the speaker’s intention during any verbal interaction be correctly recognized. This is of paramount importance especially when translating/interpreting speech communication in diplomatic discourse. This paper is a result of a contrastive study, conducted in the field of pragmatics, which aims to examine Portuguese and Croatian diplomatic discourse, focusing on expressions of gratitude and greetings in both languages. The aim of the study is to discover the linguistic processes underlying the speech acts in the two languages. This will be achieved by means of a contrastive analysis of utterances in Portuguese and Croatian. As a result of the study, some general characteristics of diplomatic discourse will be identified. The study is based on an in-depth analysis of two corpora of verbal communication, Portuguese and Croatian, within the realm of diplomatic discourse. The results of the study show that the similarities between the two linguistic systems outweigh the differences, thus allowing the author of the study to identify some general characteristics and principles underpinning expression of gratitude and greetings within diplomatic discourse. The Master’s Thesis also presents the practical implications of the study by trying to schematize the various linguistics forms through which gratitude and greetings are expressed in the discourse in question.
Prepoznavanje komunikacijske namjere govornika bitan je uvjet za uspješno obavljanje prevoditeljske aktivnosti. Važnost te komunikacijske kompetencije prevoditelja osobito se ističe u govornim situacijama koje se odvijaju o okviru diplomatskog diskursa. Ovaj rad, iz područja pragmalingvistike, je rezultat istraživanja govornih činova pozdravljanja i zahvaljivanja koji se pojavljuju u diplomatskim govorima na portugalskom i na hrvatskom jeziku. Glavni cilj istraživanja je opisati lingvističke procese koji uvjetuju realiziranje gore navedenih govornih činova. Kontrastivnom analizom načina na koji se pozdravi i zahvale ostvaruju na portugalskom i na hrvatskom jeziku, otkrivaju se i glavne značajke diplomatskog diskursa. Korpus je sastavljen od niza usmenih tekstova ostvarenih u komunikacijskoj situaciji u kojoj izvorni govornik bilo portugalskog, bilo hrvatskog jezika obavlja poslove diplomatskog predstavnika svoje države. Rezultati analize korpusa upućuju na sličnosti između dva jezična sustava prilikom ostvarivanja govornih činova pozdravljanja i zahvaljivanja u okviru diplomatskog diskursa. Stoga se može zaključiti da na jezičnu aktivnost govornika utječu čimbenici koji nadilaze jezičnu razinu.