|
|
The internal and external variation in speaker assessments based on speech performance
Nagyová, Eliška ; Volín, Jan (advisor) ; Chládková, Kateřina (referee)
This bachelor thesis deals with the internal and external variation in speaker's assessments based on speech performance. Using the material of two poems ("Portrét" by Karel Toman and "Píseň zhýralého jinocha" by František Gellner), the variability in the evaluation of the recordings of 32 speakers (16 women and 16 men) was studied. Their performance was rated by 30 listeners on a seven-point scale based on their impression of the overall recitation. The general nature of the perception test instructions allowed us to observe which factors may play an important role in the perception of the overall speech performance. The results point, among other things, to the possible influence of the text of the poems on respondents' assessments. This study serves as a basis for further in-depth research in the area of speech assessment. Keywords: internal variation, external variation, voice perception, voice attractiveness, voice measurement, voice assessment, recitation
|
|
|
Acoustic parameters of the speech signal and their intra- and interindividual variability in Czech
Houzar, Alžběta ; Skarnitzl, Radek (advisor) ; Volín, Jan (referee) ; Hes Svobodová, Marie (referee)
The doctoral thesis focuses on the variability of a wide range of acoustic parameters of the speech signal in the Czech language. It summarizes the methodological approaches applied in previous studies on the acoustic characteristics of speech and presents a comprehensive analysis of intra-speaker and inter-speaker variability of an extensive set of selected parameters through original experiments conducted using a unified speech material. The individual experiments examined acoustic parameters that characterize fundamental frequency, vowel formants, short- and long-term frequency spectrum, and the temporal domain of speech, employing a total of 61 metrics. The variability of all parameters was observed within an individual's speech during one type of utterance as well as in relation to speaking style, with recordings of both read and spontaneous speech from thirty speakers being analyzed. Based on the experimental results, information regarding the distribution of parameter values in both speaking styles within the given population was obtained. In the final experiment, a comprehensive analysis of all investigated parameters was conducted, focusing on their mutual relationships and their contribution to the overall variability of the speech signal. The results of this comprehensive analysis...
|
|
|
Information structure of utterances and the amplitude of speech signal in Czech narratives
Hapka, Haštal ; Volín, Jan (advisor) ; Šturm, Pavel (referee)
The issue of functional sentence perspective divides each utterance into a theme (a contextually inferable component known from a previous context or situation) and a rheme (a non-contextually involved component that is new to the utterance). It thus sets itself against formal syntax, which deals only with the compositional relations between the individual sentence members, and focuses on the information structure. The aim of this paper was to focus on the acoustic parameter of speech signal amplitude for both of these components separately and to see whether either of them shows higher or lower values than the other with statistical significance. Taking into account that in the flow of spoken speech speakers use different means to create prominence (stressing certain words or syllables), which affect the acoustic parameters (frequency, duration and amplitude), the hypothesis arose that the contextually uninvolved (i.e. new) information could show higher amplitude values than the contextually involved component because of the need to stress it. At the same time, during the course of each utterance, the amplitude values decrease, and the new information tends to be in the unmarked word order at the end of the utterance. On a corpus of 100 recordings from ten speakers, three methods of measurement...
|
|
|
Fundamental frequency behaviour in Czech sonorous syllabic codas
Svatošová, Michaela ; Volín, Jan (advisor) ; Skarnitzl, Radek (referee)
It has been established that listeners perceive intonation only in certain parts of the fundamental frequency (F0) contour. Descriptions of melodic patterns are often based on F0 values in vowels. However, pitch can be controlled also in sonorants (nasals, approximants), which can be part of syllabic codas in Czech. The aim of this thesis was to compare syllables with a sonorous coda and without a coda, and to examine whether the F0 contours are more similar in the domain of the syllabic nucleus or rather in the domain of the syllabic rime. The material consisted of a two-hour collection of audiobook samples recorded by sixteen professional actors. The analysis used multiple methods - representing F0 con- tours by selected discrete points, modelling the whole F0 contour with Legendre polynomials, and functional principal component analysis (FPCA). The results indicated that the syllabic rime represents the domain relevant for the production of functional melodic patterns (melodemes) in Czech. F0 contours in syllabic rimes showed more similar values of span (across both types of sylla- bles) than F0 contours measured in syllabic nuclei. Their shapes were also more comparable when syllabic rimes were considered. In syllables with a sonorous coda, the relationship between the slope of the F0 contour...
|
|
|
Investigating prosody in spoken Czech: A corpus-linguistic approach
Lukeš, David ; Vondřička, Pavel (advisor) ; Volín, Jan (referee) ; Čech, Radek (referee)
Prosody is a key aspect of spoken language, yet it is currently underrepresented in the spoken Czech corpora on offer at the Czech National Corpus. This is mainly because spoken corpora are very expensive and manual work intensive as it is, and adding more annotation manually is infeasible. The present dissertation thus charts a way to provide an automatic prosodic annotation for the spoken corpora of the CNC using the Prosogram framework, in combination with other tools and various custom postprocessing strategies and heuristics. Acaseisalsomadeinfavoroftheory-light,predominantlydescriptiveapproaches when preparing general-purpose spoken corpus annotations for the consumption of the linguistics research community at large, in a variety of contexts and research tasks. This case is philosophically anchored in a discriminative approach to meaning, which is shown to be the correct, paradox-free alternative to the currently more dominant paradigm of compositionality. Finally, a selection of results based on the Prosogram-generated annotation is presented. A particular focus is given to pitch range, which is characteristically restricted in Czech compared to other languages like English, but other features such as glissandos are also considered. Keywords: Czech, speech, prosody, corpus linguistics,...
|
|
|
Forensic potential of temporal structure metrics in Czech spoken texts
Jalová, Marie Zoe ; Volín, Jan (advisor) ; Skarnitzl, Radek (referee)
This thesis focuses on temporal structure metrics in a speech. The theoretical part describes their character and their possibility in forensic use, the practical part examines, based on two recordings of poetry, the behavior of temporal structure metrics in a relation to the material and the identity of a speaker. This thesis makes a contribution to the description of temporal structure of the czech language, specifically to the poetry genre, with a potential use in forensic phonetics.
|
|
|
Canonical and actual speech and articulation rate throughout spoken dialogues
Línková, Alena ; Skarnitzl, Radek (advisor) ; Volín, Jan (referee)
Speech tempo and its variability in Czech have been the subject of many studies, but most of them have primarily focused on monologues. In this bachelor thesis, we focus on speech tempo variability in dialogues, as well as on different ways of measuring speech tempo and how that can affect the values obtained. The thesis includes a theoretical section that summarizes current knowledge regarding speech tempo, different ways of objectively measuring it and factors affecting its values, perception of speech tempo and factors that affect it, pauses, and the specifics of speech tempo in dialogues. In the practical section, we analysed the recordings of 13 speakers participating in a television political debate. We measured articulation rate in prosodic phrases and speech rate in clauses and thematically unified paragraphs, as well as local rate in the initial, medial, and final thirds of these dimensions. We measured syllable (syll/s) and phone rate (ph/s), and canonical and surface (syllable) rate. We found statistically significant differences between tempo values in thirds of clauses (syll/s) and prosodic phrases (ph/s), but not in paragraphs (syll/s). The effect of the length of a prosodic phrase on both syllable and phone rate values was, however, found not to be statistically significant. Powered by TCPDF...
|
|
|
Phonetic structure of prosodic phrases in TV talk shows in Czech and American English
Hladká, Adéla ; Volín, Jan (advisor) ; Chládková, Kateřina (referee)
The main aim of this bachelor thesis was to describe prosodic phrasing in the TV debate genre in the context of comparing Czech with American English. The aim was to find similarities and differences between the two languages/cultures in the area of phrasing and to see if inter- individual or intra-individual differences are greater than the differences between the situation in Czech and American English. 4 different sets of recordings of televised political debates (Nedělní partie and Arizona Horizon's), 2 Czech and 2 English, were analyzed, with a total of 14 speakers, two of whom, the moderators, were repeated in the same debate, bringing the total number of speakers to 12. The data for the analysis were provided by the Phonetic Institute of the Faculty of Arts of Charles University under the COCOSDA principle. For each speaker, the number of words, syllables and vowels per phonemic clause was analysed. To extract the resulting data, I used scripts from the phonetic software Praat. The shortest and longest phrases were also examined in an attempt to figure out if these phenomena are somehow generalizable in this genre of conversational speech. The results of this research show that although speakers differ within language groups, the personality of some individuals and the environment in which...
|
|
|
Perception of affective states in speech in variable musical context
Englmaierová, Martina ; Bořil, Tomáš (advisor) ; Volín, Jan (referee)
This thesis deals with perception of affective states in speech in variable musical context. We assembled a perception test from 15 - 20s excerpts from Czech films, which contained anger or joy in speech. Then we added different types of music to the excerpts according to Russell's model of affect. Respondents were asked to assess the way the items affected them on the scales of arousal and pleasantness. The test consisted of a total of 74 items in four blocks: 24 test items (4 items with joy and 4 with anger in speech in variants without music, with music supporting and with music not supporting the affect in speech) and 50 distractors in pseudo- random order. The test was performed by 56 respondents. We have found that joy is not as dominant as anger in speech, which is very unpleasant and arousing. However, any music has increased the arousal and pleasantness of joy. Despite Russell's model, sadness also caused more arousal in relation both to joyful and angry items. In accordance with the model, relaxed music increased the pleasantness of joy while angry and sad music decreased it. Supportive music emphasized even more pleasantness and arousal in the angry affect in speech. Contrary to expectations, only relaxed music (not joyful) increased the pleasantness. However, arousal was also increased...
|
|
|
Personal characteristics of speech rhythm in Russian newsreading
Čížková, Irena ; Volín, Jan (advisor) ; Veroňková, Jitka (referee)
Personal characteristics of speech rhythm in Russian newsreading Bc. Irena Čížková Abstract An issue of individual rhythmic characteristics of particular 5 native speakers of the Russian language, newsreaders from the BBC, is described in this thesis. A research of the given 5 speech recordings was conducted based on the academic concepts created by Ramus, Mehler and Nespor, and by Low and Grabe and also by Dellwo, so through a speech rhythm research based on rhythm correlates that are related to vocalic and intervocalic intervals. The recordings were processed in an analyting program called Praat and the extracted results were then evaluated in a statistical processor called STATISTICA. These materials were used for further analysis. The recordings were analysed based on several parameters: %V (proportion of vocalic intervals in one breath group), ∆C and ∆V (standard deviation of the vocalic and consonantal interval duration), PVI-V, PVI-C (Pairwise Variability Index of the vocalic and consonantal interval duration), Varco V and Varco C (variation coefficient of the vocalic and consonantal interval duration) and the difference between duration of stressed and unstressed vowels. Three parameters that were the most successful from the speakers' ability to differentiate point of view were selected through an...
|
guest ::
