Data-Driven Pronunciation Generation for ASR Maria Obedkova In ASR systems, dictionaries are usually used to describe pronunciations of words in a language. These dictionaries are typically hand-crafted by linguists. One of the most significant drawbacks of dictionaries created this way is that linguistically motivated pronunciations are not necessarily the optimal ones for ASR. The goal of this research was to explore approaches of data-driven pro- nunciation generation for ASR. We investigated several approaches of lexicon generation and implemented the completely new data-driven solution based on the pronunciation clustering. We proposed an approach for feature extraction and researched different unsupervised methods for pronunciation clustering. We evaluated the proposed approach and compared it with the current hand-crafted dictionary. The proposed data-driven approach could beat the established base- lines but underperformed in comparison to the hand-crafted dictionary which could be due to unsatisfactory features extracted from data or insufficient fine tuning. 1

Die vorliegende Arbeit beschäftigt sich mit den vokalischen und konsonantischen Lautschwächungen in den Dienstwörtern des Deutschen. Die Aktualität solcher Arbeit besteht darin, dass diese Gruppe von Wörtern lautlich besonders stark abgeschwächt wird und zugleich das Typische eines Gesprächstextes darstellt, obwohl die Dienstwörter sehr wenig zur eigentlichen Sachverhaltsdarstellung beitragen, denn sie enthalten einen nur sehr geringen Informationsgehalt. Der theretische Teil erläutert das Wesen und Merkmale der Dienstwörter, ihre phonetischen Besonderheiten und Erscheinung der Lautschwächungen in der Wissenschaft, ihr Auftreten in den Dienstwörtern in der modernen deutschen Sprache. Im praktischen Teil wird es mit Hilfe von zwei Texten aus verschiedenen Stilen erforscht, wie sich die Vokal- und Konsonantenschwächungen in den Dienstwörtern in verschiedenen phonetischen Stilen äußern.

<p>Predmet doktorske disertacije je analiza fonetskih i morfolo&scaron;kih gre&scaron;aka načinjenih prilikom učenja rumunskog jezika kao stranog jezika na Univerzitetu u Novom Sadu od &scaron;kolske 2010/11. do &scaron;kolske 2015/16. godine na nivoima A1 i A2. Korpus čine testovi zavr&scaron;nih ispita, kolokvijuma i usmenih delova ispita 159 studenata sa različitih studijskih grupa. Cilj rada je sistematizovanje gre&scaron;aka prema fonetskim i morfolo&scaron;kim nivoima i tipovima i opisivanje gre&scaron;aka prema uzrocima njihovog nastajanja.<br />Rad je podeljen na tri glavne celine. Prva celina sadrži predmet, cilj i zadatke istraživanja, zatim metode i tehnike istraživanja, kao i izvore za korpus. Druga celina započinje teorijskim okvirom, a nastavlja se pregledom postojeće relevantne literature o učenju stranog jezika, razmatranjem razloga za učenje stranog jezika, kao i uloge profesora u procesu učenja. Posebno su predstavljene gre&scaron;ke u učenju stranih jezika. Treća celina odnosi se na samu analizu dobijene građe, gre&scaron;aka i interpretaciju dobijenih rezultata. Ova celina podeljena je na dva veća poglavlja koja se bave fonetskom i morfolo&scaron;kom analizom gre&scaron;aka. U okviru poglavlja koje se bavi fonetskim gre&scaron;kama obrađene su sledeće oblasti: samoglasnici, diftonzi, triftonzi, alternacija samoglasnika, suglasnici, glasovne grupe ge, gi, ghe, ghi, ci, ce, chi, che, kao i alternacija suglasnika. U okviru drugog poglavlja, koje se bavi morfolo&scaron;kim gre&scaron;kama, obrađene su sledeće oblasti: pogre&scaron;na upotreba imenica, pogre&scaron;na upotreba određenog i neodređenog člana, gre&scaron;ke u upotrebi zamenica (lične zamenice u nominativu i akuzativu, povratno-refleksivne zamenice, pokazne zamenice), gre&scaron;ke u upotrebi prideva i gre&scaron;ke u upotrebi brojeva. U pomenutom poglavlju obrađena su i glagolska vremena (prezent, perfekat, futur I i kolokvijalni futur), kao i lični (konjunktiv i imperativ) i nelični glagolski načini (infinitiv i particip sa aktivnim značenjem). Na kraju poglavlja obrađeni su i predlozi (la i &icirc;n). Budući da su nastavni proces pratile promene koje se odnose na primenu modernih tehnologija u učenju stranog jezika i otklanjanju gre&scaron;aka, deo na&scaron;eg istraživanja je posvećen i tome.<br />Rezultati analize pokazali su da je najveći broj identifikovanih gre&scaron;aka zapravo na morfolo&scaron;kom nivou, ali da fonetske gre&scaron;ke predstavljaju prepreku u učenju rumunskog jezika. U najvećem broju primera ustanovili smo da su gre&scaron;ke unutarjezičkog karaktera, ali i da su studenti primenjivali strategije transfera iz maternjeg ili nekog drugog stranog jezika koji znaju. Namera nam je bila da ukažemo na koje gramatičke kategorije profesori, ali i studenti, moraju da obrate pažnju prilikom učenja rumunskog kao stranog jezika i koji su nedostaci nastavnog procesa. Rezultati istraživanja će se koristiti u nastavi rumunskog jezika kao stranog i<br />Fonetska i morfolo&scaron;ka analiza gre&scaron;aka u učenju<br />rumunskog jezika kao stranog jezika<br />mogu pomoći u kreiranju posebnog kurikuluma. Gre&scaron;ke se u dobroj meri mogu eliminisati kori&scaron;ćenjem posebnih metoda i vežbi, kao i inovacija u procesu nastave, jer je motivacija za učenje jezika mnogo veća.</p>

While handling the task of speaker identification, forensic phoneticians use a combination of various parameters contained in different levels of speech signal. The main aim of the present thesis is to explore whether glottalization in Czech may be considered as a potentially useful parameter in this sense. In our research, we focus on the rate of prevocalic glottalization at word boundaries and we distinguish between different realisations of glottalization: canonical glottal stop and its hypoarticulated form - creaky voice. The studied material consists of repeated recordings of three male and four female speakers and contains both read text and spontaneous speech. The results do not indicate that the same speaker would use glottalization differently in the first and second recording, but a difference in glottalization is found between speakers. From the forensic phonetics point of view, this finding seems to be useful. Marginally, some other factors which are not directly connected with the speaker (height of following vowel, lexical factors and speech rate) were examined, but no influence on glottalization was found. Keywords: glottal stop, glottalization, forensic phonetics, speaker identification

TITLE: Distinctive Features of Irish Accent AUTOR: Tomáš Uher DEPARTMENT: Department of English Language and Literature SUPERVISOR: Mgr. Kristýna Poesová ABSTRACT: The aim of this thesis is to examine the distinctive phonetic and phonological aspects of Irish English. The theoretical part provides a comprehensive description of the segmental and suprasegmental features of the Irish accent in contrast with the Received Pronunciation. The first section of the practical part analyses the speech of three Irish actors in Irish films and TV shows in order to examine the use of the distinctive features described in the theoretical part. The second section of the practical part presents a 45 minute long lesson plan on the Irish accent with suggested exercises for intermediate (B2) learners. KEY WORDS: accent, Irish, pronunciation, phonetics, phonology, lesson plan

(in English): Voice disguise can potentially occur in every utterance that is associated with any criminal case. In order to identify the perpetrator it is necessary to analyze the speech and understand how the different types of voice disguise can affect the speaker's voice qualities. This thesis focuses on the ability of voice disguise, portraying three groups of speakers in relation to their educational background. The aim of this work is to determine the strategies adopted by the speaker to conceal his/her identity and furthermore it poses the question whether differences among the three groups of speakers, their choice of strategy and its inherent success can be found. The basis for this research stems from 86 recordings which were undertaken in Pilsen and Prague with 43 young people aged 20 to 31. Two read utterances, one undisguised and the other freely disguised, were obtained from each of the participants and were compared with each other. The results show that the preferred forms of voice disguise appeared to involve changes in phonation - especially decrease or increase of fundamental frequency of the speaker's voice. Among the three groups of speakers, their choice and the success of the chosen strategy only minor differences could be found, yet for a final confirmation of this...

The dynamics of Southern Carpathian dialects - abstract The aim of this work is to describe the development of Southern Carpathian dialects in two state entities, in northeastern Slovakia and the Transcarpathian region of Ukraine, in the last 70 years. Starting material are recordings of dialectal speech and their transcription obtained in own field research conducted since 2011 within the studied area, especially in villages located close to the Slovak-Ukrainian state border. The focus of this work is a description of the dynamics of the phonetic, morphological and lexical system of the dialects of two villages, Ublya, which was chosen due to the long tradition of dialect research, and Malyi Bereznyi, the nearest neighboring village on the territory of Ukraine. Here until World War II closely related Southern Carpathian dialects had been developing in the same sociolinguistic conditions. After World War II they were separated by the Czechoslovak- Soviet border and came under the influence of different standard languages - Slovak, on the one hand, and Ukrainian and Russian on the other hand. A comparison of older material with the results of our field research enables to observe a dialectal divergence caused by leveling processes oriented to different standard languages. Data obtained in Ublya and Malyi...

This thesis deals with the description of the human voice from the forensic phonetic perspective. At the beginning of the theoretical part, some important cases from the history of this field are mentioned, as well as the attitude of judges to the voice as evidence.The main tasks of forensic phonetics are then briefly presented. The remainder of the theoretical part is devoted to summarizing the existing knowledge from the field of audiovisual perception and comparison of voice description and face description, including several concrete examples of systematic approaches to obtaining these descriptions, both for professional purposes and for the needs of investigators in practise. The aim of the practical part is to compare the initial uninstructed, naive description of the voice of selected speakers, obtained from the respondents in a simulated police questioning, followed by an instructed systematic description according to the interrogation protocol, which uses the layman's formulation of the individual questions instead of the original phonetic terminology, and illustrative sound samples to make it easier for the respondents to understand the protocol and to give a more detailed account of the heard voice. The first part of the results analysis is devoted primarily to the naive testimony of the...

Speech contains certain attributes characteristic for a speaker, so-called idiosyncratic features. This thesis focuses on the form of these features in intentional voice disguise - whether speakers are able to change them in a substantial way, or if they tend to remain stable in spite of intentional speech modifications. It was also investigated whether any general tendencies to similar changes of such features under voice disguise exist among the speakers. The observed features were statistical f0 indicators, f0 contours, vowel formants, long- term formant distributions, spectral characteristics of sibilants, intensity, intensity contours, speech and articulation rate, %V and local articulation rate contours. In f0 median and standard deviation, vowel formants, LTFDs, intensity, articulation rate, and %V, prominent shifts under voice disguise were observed in general; in the majority of these parameters, the shifts differed among speakers. However it was found that the value of %V generally tends to rise under voice disguise. Also, intensity showed an increase in majority of cases. In f0 contours, similar patterns were observed among speakers in normal speech, however, in disguised speech, greater differences appeared among speakers; speakers tend to employ nonstandard dynamic f0 patterns more...

Ce travail montre la variation socio-phonétique de la vibrante /r/ dans la province basque de Soule. La prononciation de cette consonne est en constante évolution et le dialecte souletin ne fait pas exception. Cette évolution a été étudiée suivant la variation diastratique qui nous amène à étudier les changements par le biais de facteurs sociaux.En basque et ici en souletin, il y a deux différentes vibrantes : d’un côté il y a la vibrante battue /ɾ/ et de l’autre la roulée /r/. La première vibrante ne se prononce généralement pas entre deux voyelles en souletin. La seconde en revanche se prononce dans tous les contextes. Mais ces vibrantes connaissent à ce jour deux formes de prononciation à savoir la forme apicale (/r/) et la forme uvulaire (/ʀ/). C’est pour cela que différentes options s’offrent à chaque interlocuteur souletin. Nous verrons dans ce travail que les choix ce font par le biais des facteurs sociaux. Les jeunes ont tendance à utiliser un son plutôt qu’un autre et les plus âgés en utilisent d’autres.Les données ont été recueillies grâce à des interviews effectuées sur le terrain grâce à un questionnaire basé sur la traduction de mots. Il leur a été demandé de traduire les même mots afin d’avoir une base de donnée qui permettrait par la suite de comparer les vibrantes dans des contextes différents. Les interlocuteurs ont été choisis suivant leur âge (jeunes, adultes, personnes plus âgées), leur sexe (homme ou femme), l’origine de leur basque (si ils l’ont appris à la maison ou à l’école), la langue dans laquelle ils travaillent (en basque ou en français) et si ils ont quitté le territoire du Pays Basque plus de 4 mois ou non.Ces données ont ensuite été analysées principalement grâce à deux programmes qui sont Praat et SPSS. C’est avec ces résultats recueillis que des recherches statistiques ont été menées, et que nous avons pu en ressortir l’état actuel des vibrantes en Soule. / This contribution shows the variation of the pronunciation of the /r/ in Zuberotar, an Eastern dialect of the Basque language. The pronunciation of this consonant is changing and depends on different social factors. There are two consonants: alveolar tap /ɾ/ and alveolar trill /r/. While the first consonant can be pronounced or can disappear between vowels (hari vs hai 'wire'), the second one can be pronounced as a vibrant or trill (/r/) or as a fricative uvular (/ʀ/). Therefore, people who speak this dialect have different options: no pronunciation of the consonant, pronunciation as alveolar tap, as an alveolar trill or as an uvular. These options depend on social factors surrounding the speaker or on the features of the words such as its origin and the adaptation into dialect structure. For instance, in the word euro, which is borrowed from French «r» is always pronounced Data for the research was recorded during interviews led by a questionnaire. The participants were asked to produce the same sentences. They were selected according to the age (young / adults / older), sex (male and female), native or familiar tongue (Basque vs. French), language used in professional context and if they lived abroad for more than 4 months... Data have been analyzed linguistically and statistically using PRAAT and SPSS software. In this way, the results will supply the reader with the direction of the variation of this consonant in Zuberotar. / Euskaraz eta hemen zubereraz, bi dardarkari mota desberdin badira : /ɾ/ bakunak eta /r/ anizkunak. Bakunaren kasuan, edo ahoskatzen da edo desagertzen da bi bokal artean /hari/ edo /hai/ kasuetan bezala. Anizkunaren kasuan aldiz, bi motatakoak badira : apikariak /r/ edo ubularrak /ʁ/. Horretarako, informateek ahoskatze hautu desberdinak badituzte. Lan horretan ikusiko da faktore sozialen bidez hautuak eginak direla, hala nola gazteek egiten dituzten hautuak ez dira baitezpada zaharrek egiten dituztenak.Dardarkari horien guztien aztertzeko, inkestak egin ditugu. Inkesta hori bideratua izan da itzulpenen bidez eramana izan baita. Honetan, hitz zenbaitzuen itzultzea galdegina izan zaie. Aldi bakoitzean ber hitzak izan dira ondotik konparaketa baten eramateko. Hiztunak haien adinaren arabera hautatuak izan dira (gazte, heldu, zahar), generoaren arabera (emazte, gizon), haien euskararen jatorria (euskara etxean edo eskolan ikasi dutenez), lanean erabiltzen duten hizkuntza (euskara edo frantsesa), edo Euskal Herritik kanpo lau hilabete baino gehiago joan direnez.Ondotik, nagusiki bi programekin lan egin da: Praat eta SPSS programak. Eta programa horiei esker bildu diren datuekin, azterketa estatistikoak eraman dira eta gaur egungo dardarkarien argazki bat atera ahal izan da.