Submitted by Kamila Costa (kamilavasconceloscosta@gmail.com) on 2015-06-16T21:49:25Z No. of bitstreams: 1 Dissertação-Jorge L de F Lima.pdf: 1385009 bytes, checksum: 750a3974656cc69574b553858a336cb8 (MD5) / Approved for entry into archive by Divisão de Documentação/BC Biblioteca Central (ddbc@ufam.edu.br) on 2015-06-17T15:36:27Z (GMT) No. of bitstreams: 1 Dissertação-Jorge L de F Lima.pdf: 1385009 bytes, checksum: 750a3974656cc69574b553858a336cb8 (MD5) / Approved for entry into archive by Divisão de Documentação/BC Biblioteca Central (ddbc@ufam.edu.br) on 2015-06-17T15:38:32Z (GMT) No. of bitstreams: 1 Dissertação-Jorge L de F Lima.pdf: 1385009 bytes, checksum: 750a3974656cc69574b553858a336cb8 (MD5) / Made available in DSpace on 2015-06-17T15:38:32Z (GMT). No. of bitstreams: 1 Dissertação-Jorge L de F Lima.pdf: 1385009 bytes, checksum: 750a3974656cc69574b553858a336cb8 (MD5) Previous issue date: 2014-09-29 / Não Informada / People use various means of communication at all times during the length of their lives, however, do not ask how this process. However, it is known that how individuals speak and the differentiation of which they speak, occur according to the culture in which the individuals in question are inserted. To better illustrate this situation, the work here this shows in its content the semiotic field of the Fair city of Benjamin Constant (MA) - city inhabited by various indigenous ethnic groups, as well as Peru and clear, Brazilians - with the objective of investigating the cultural influences on the use of the orality by agents involved in trade relations in the fair city of Benjamin Constant-AM and the implication of this for the understanding of how does the communicative process in a frontier region, having as a basis for theoretical study authors as: Bordieu (1999); de Certeau (2012), Geertz (1989); Hall (1997); Machado (2007); Matos (2001) and Martins (2002). From this perspective, we used the ethnography as a method of research, which enabled us show by means of interviews and participant observation that the language influences the life of human beings, was how important it is and that, even with many varieties existing language on the planet, there is always the possibility of communication and that by virtue of contacts in border region - several possibilities of communication can be "engendered". We noted in the fair city of Benjamin Constant when to overcome communicative difficulties resulting from the different languages in contact, replaced the orality of bodily practices. / As pessoas utilizam diversos meios de comunicação a todo o momento durante a extensão de suas vidas, no entanto, não se perguntam como se dá esse processo. Sabe-se, porém, que a forma como os indivíduos falam e a diferenciação de como falam, ocorrem de acordo com a cultura na qual os indivíduos em questão estão inseridos. Para melhor ilustrar essa situação, o trabalho aqui presente apresenta em seu conteúdo o campo semiótico da Feira Municipal da cidade de Benjamin Constant (MA) – cidade habitada por diversas etnias indígenas, além de peruanos e claro, brasileiros - com o objetivo de Investigar as influências socioculturais no uso da oralidade pelos agentes envolvidos nas relações comerciais na feira municipal de Benjamin Constant-AM e a implicação disso para a compreensão de como se caracteriza o processo comunicativo numa região de fronteira, tendo como base de estudo teórico autores como: Bordieu (2009); Certeau (2012); Geertz (1989); Hall (1997); Machado (2007); Matos (2001) e Martins (2002). Nessa perspectiva, utilizou-se a etnografia enquanto método de investigação, o que nos possibilitou mostrar por meio de depoimentos e observação participativa que a linguagem influencia a vida dos seres humanos, mostrou-se o quão importante ela é e que, mesmo com muitas variedades linguísticas existentes no planeta, há sempre a possibilidade de comunicação e que por força dos contatos em região de fronteira - possibilidades diversas de comunicação podem ser “engendradas”. O que se constatou na feira municipal de Benjamin Constant quando para superar as dificuldades comunicativas resultantes das diferentes línguas em contato, substituiu-se a oralidade por práticas corporais.

The application of the linear prediction Model for speech waveform analysis to context-independent automatic speaker recognition is explored, primarily in terns of the parametric sensitivity of the model. Feature vectors to characterize speakers are formed from linear prediction speech parameters computed as inverse filter coefficients, reflection coefficients or cepstral coefficients, and also power spectrum parameters via Fast Fourier Transform coefficients. The comparative performance of these parameters is investigated in speaker recognition experiments. The stability of the linear prediction parameters is tested over a range of model order from p=6 to p=30. Two independent speech databases are used to substantiate the experimental results. The quality of the automatic recognition technique is assessed in a novel experiment based on a direct performance comparison with the human skill of aural recognition. Correlation is sought between the performance of the aural and automatic recognition methods, for each of the four parameter sets. Although the recognition accuracy of the automatic system is superior to that of the direct aural technique, the error distributions are highly variable. The performance of the automatic system is shown to be empirically based and unlike the intuitive human process. An extended preamble to the description of the experiments reviews the current art of automatic speaker recognition, with a critical consideration of the performance of linear prediction techniques. As supported by our experimental results, it is concluded that success in the laboratory rests upon a rather fragile foundation. Application to problems beyond the controlled laboratory environment is seen, therefore, to be still more precarious.

In this thesis, the speaker and emotion recognition system is established by PC and digit signal processor (DSP). Most speaker and emotion recognition systems are separately accomplished, but not combined together in the same system. In this thesis, it will show how speaker and emotion recognition systems are combined in the same system. In this system, the voice is picked up by a mike and through DSP to extract the characteristics. Then it passes the sample correctly, it can draw the result of distinguishing. The recognition system is divided into four sub-systems: the pronunciation pre-process, the speaker training model, the speaker and emotion recognition, and the speaker confirmation. The pronunciation pre-process uses the mike to capture the voice, and through the DSP board to convey the voice to the SRAM, then movements dealt with pre-process. The speaker trained model uses the Gaussian mixture model to establish the average, coefficient of variation and weight value of the person who sets up speaker specifically. And we¡¦ll take this information to be the datum of the whole recognition system. The speaker recognition mainly uses the density of probability to recognition the speaker¡¦s identity. The emotion recognition takes advantage of the coefficient of variation to recognize the emotion. The speaker confirms is set up to sure whether the user is the same speaker who hits for the systematic database. The recognition system based on DSP includes two parts¡GHardware setting and implementation of speaker algorithm. We use the fixed-arithmetician DSP chip (chipboard) in the DSP, the algorithm of recognition is Gaussian mixture model. In addition, compared with floating point, the fixed point DSP cost much less; it makes the system nearer to users.

There are a lot of difficulties that have to be overcome in the speaker-independent (S.I.) phrase recognition system . And the feasibility of accurate ,real-time and robust system pose of the greatest challenges in the system. In this thesis ,the speaker-independent phase recognition system is based on Hidden Markov Model (HMM). HMM has been proved to be of great value in many applications, notably in speech recognition. HMM is a stochastic approach which characterizes many of the variability in speech signal. It applys the state-of-the-art approach to Automatic Speech Recognition .

The increasing number of proficient, non-native English speakers, both in U.S. academic institutions and around the globe, warrants considerable investigation into possible norms developing within non-native to non-native interactions. This report analyzes attitudes toward accent, a prominent indicator of foreignness, within non-native English speaker interactions. It presents relevant research on this topic, and it summarizes some of the major findings of an online survey that examined what attitudes, if any, non-native listeners may form on the basis of accent alone when listening to other non-native English speakers. The results suggest that listeners base attitude judgments more on native-likeness than on intelligibility. Also, speakers’ perceptions of their own non-native accent are more negative than how they actually rate themselves as compared to others. / text

The present study examines the acoustic and articulatory features of English onset and coda /l/s as produced by native English and Mandarin Chinese speakers in the vowel contexts of /i/, /ɪ/, /e/, / ɛ/, /u/, /ʊ/, /o/, /ɔ/, /ɑ/, /ʌ/, /ɚ/, and /æ/, and via the elicitation tasks of word list and mini dialogue. Four Mandarin Chinese speakers who had lived in Canada for at least one year by the time of the experiment and four Canadian English speakers who were born and raised on west coast of Canada participated in the research. Both groups of speakers were the graduate students studying at the University of Victoria. The experiment took place at the Phonetics Laboratory in the Department of Linguistics at the University of Victoria. An ultrasound machine together with a synchronized microphone was used to record the speech data for analysis. The results showed that for onset /l/, the tongue position of the Mandarin Chinese speakers was more front than that of the English speakers. For coda /l/s, Mandarin Chinese speakers had lower and more retracted tongue position than their English counterparts. ANOVA tests showed that vowel contexts and task formality had limited impact on the acoustic qualities of the onset and coda /l/s produced by both groups of speakers. The results and conclusions from the present study will contribute to a better understanding of the articulatory features of the English /l/s. Mandarin Chinese learners may also benefit from this study in that they could potentially improve their pronunciations and reduce accent. / Graduate

Actuating a cantilever piezoelectric element over a frequency spectrum, the movement will show resonances and hysteresis behavior not present in the input signal. Excursion modeling and open loop control of a cantilever piezoelectric bimorph actuator was studied in this thesis, with the aim to enhance the actuator's movement to more accurately render audible input. This actuator has lower energy consumption and presents new possibilities for speaker design in constrained situations compared to conventional micro speaker technology. Much work has previously been done to model piezoelectric cantilever actuators below the first and second resonance frequency. This thesis describes a physical linear model and its modifications to render the eight first resonance frequencies below 20 kHz, as well as the model's performance in open loop control. This was performed on a single piezoelectric beam and a concept piezoelectric speaker. For the single piezoelectric beam the model was validated with fair overall result below 3 kHz. The model is suggested to have fair overall behavior up to 15 kHz. Above 15 kHz the experiments showed changed characteristics that were not modeled well. The open loop control had the intended behavior but severe resonances and physical constraints made the open loop control ineffective to enhance the sound rendering. Two different approaches were used for trying to improve the sound rendering based on an excursion model. These approaches did not generate useful methods but present viable input to future work with this type of speaker structure, for reducing disharmonics and creating a physical design tool for sound simulation. For the concept piezoelectric speaker, due to difficulties in measuring excursion, the model could not be validated. This made the approaches for enhancing the sound rendering ineffective. However, it can be concluded from the concept speaker that the cantilever piezoelectric speaker technology has qualities that could compete with the conventional micro speaker technology. Challenges remain in electric hardware, actuator configuration and acoustic design as well as in fine tuned signal processing for the concept speaker to become a competitive product.

The present paper aims at disclosing what metalinguistic elements of language are used in dialogues and what are the intended meanings expressed by speakers using particular metalinguistic elements. The purpose of the research was to disclose how the metalinguistic elements of language add to the formation of a successful communicative act.The main objectives of the research were to analyse what the speaker intends to say in a communicative act by using certain words in a particular situation, and to reveal how the speaker produces a logical utterance using particular metalinguistic elements. The method used in the research was content analysis. The use of metalinguistic elements highly depends on the context of the situation and the relations between the speakers. The results of the research proved the hypothesis that metalinguistic elements of language are usually used in dialogues as only using these language elements the speaker can achieve the main aim of a communicative act,i.e. express his/her attitudes, opinions and feelings. The research proved that metalinguistic elements used in dialogues make a communicative act sound natural and lively.The use of metalinguistic elements also assures the logical sequence of a conversation. / Šis darbas siekia atskleisti kokie metalingvistiniai elementai dažniausiai naudojami dialoguose ir kokias reikšmes kalbėtojai perduoda klausytojams vartodami tam tikrus metalingvistinius elementus. Pagrindinis tyrimo tikslas buvo išsiaiškinti kaip metalingvistiniai elementai prisideda prie komunikatyvinio akto sėkmingo sukūrimo.Pagrindiniai darbo uždaviniai buvo atskleisti ką kalbantysis nori pasakyti kalbėjimo akte vartodamas tam tikrus žodžius tam tikroje situacijoje ir atskleisti kaip kalbantysis sukuria aiškų, logišką ir tikslų pasakymą vartodamas kalboje metalingvistinius elementus. Pagrindinis metodas naudojamas šiame darbe buvo turinio analizė. Metalingvistinių elementų vartojimas priklauso nuo pokalbio situacijos ir santykių tarp pokalbio dalyvių. Rezultatai gauti tyrimo metu patvirtino hipotezę kad metalingvistiniai elementai yra dažniausiai vartojami dialoge, nes tik vartodamas šiuos elementus kalbėtojas gali pasiekti pagrindinį komunikacijos tikslą- išreikšti savo nuomonę, mintis ir jausmus tam tikrų dalykų ar žmonių atžvilgiu. Tyrimas parodė, kad metalingvistiniai elementai, vartojami dialoguose suteikia komunikacijai gyvumo ir natūralumo, be šių elementų pokalbis nebūtų vaizdingas, nes kalbėtojui būtų sunkiau išreikšti savo jausmus,nuomonę ir požiūrius į tam tikrus dalykus.Metalingvistiniai elementai taip pat užtikrina logišką, rišlią pokalbio seką ir vystymą, kurie leidžia pokalbio dalyviams lengvai įsitraukti į pokalbį ir jį tęsti bei plėtoti.

Identity verification or biometric recognition systems play an important role in our daily lives. Applications include Automatic Teller Machines (ATM), banking and share information retrieval, and personal verification for credit cards. Among the biometric techniques, authentication of speakers by his/her voice is of great importance, since it employs a non-invasive approach and is the only available modality in many applications. However,the performance of Automatic Speaker Verification (ASV) systems degrades significantly under adverse conditions which cause recordings from the same speaker to be different.The objective of this research is to investigate and develop robust techniques for performing automatic speaker recognition over various channel conditions, such as telephony and recorded microphone speech. This research is shown to improve the robustness of ASV systems in three main areas of feature extraction, speaker modelling and score normalization. At the feature level, a new set of dynamic features, termed Delta Cepstral Energy (DCE) is proposed, instead of traditional delta cepstra, which not only greatly reduces thedimensionality of the feature vector compared with delta and delta-delta cepstra, but is also shown to provide the same performance for matched testing and training conditions on TIMIT and a subset of the NIST 2002 dataset. The concept of speaker entropy, which conveys the information contained in a speaker's speech based on the extracted features, facilitates comparative evaluation of the proposed methods. In addition, Frequency Modulation features are combined in a complementary manner with the Mel Frequency CepstralCoefficients (MFCCs) to improve the performance of the ASV system under channel variability of various types. The proposed fused system shows a relative reduction of up to 23% in Equal Error Rate (EER) over the MFCC-based system when evaluated on the NIST 2008 dataset. Currently, the main challenge in speaker modelling is channel variability across different sessions. A recent approach to channel compensation, based on Support Vector Machines (SVM) is Nuisance Attribute Projection (NAP). The proposed multi-component approach to NAP, attempts to compensate for the main sources of inter-session variations through an additional optimization criteria, to allow more accurate estimates of the most dominant channel artefacts and to improve the system performance under mismatched training and test conditions. Another major issue in speaker recognition is that the variability of score distributions due to incompletely modelled regions of the feature space can produce segments of the test speech that are poorly matched to the claimed speaker model. A segment selection technique in score normalization is proposed that relies only on discriminative and reliable segments of the test utterance to verify the speaker. This approach is particularly useful in noisy conditions where using speech activity detection is not reliable at the feature level. Another source of score variability comes from the fact that not all phonemes are equally discriminative. To address this, a new score re-weighting technique is applied to likelihood values based on the discriminative level of each Gaussian component, i.e. each particular region of the feature space. It is found that a limited number of Gaussian mixtures, herein termed discriminative components are responsible for the overall performance, and that inclusion of the other non-discriminative components may only degrade the system performance.

Identity verification or biometric recognition systems play an important role in our daily lives. Applications include Automatic Teller Machines (ATM), banking and share information retrieval, and personal verification for credit cards. Among the biometric techniques, authentication of speakers by his/her voice is of great importance, since it employs a non-invasive approach and is the only available modality in many applications. However,the performance of Automatic Speaker Verification (ASV) systems degrades significantly under adverse conditions which cause recordings from the same speaker to be different.The objective of this research is to investigate and develop robust techniques for performing automatic speaker recognition over various channel conditions, such as telephony and recorded microphone speech. This research is shown to improve the robustness of ASV systems in three main areas of feature extraction, speaker modelling and score normalization. At the feature level, a new set of dynamic features, termed Delta Cepstral Energy (DCE) is proposed, instead of traditional delta cepstra, which not only greatly reduces thedimensionality of the feature vector compared with delta and delta-delta cepstra, but is also shown to provide the same performance for matched testing and training conditions on TIMIT and a subset of the NIST 2002 dataset. The concept of speaker entropy, which conveys the information contained in a speaker's speech based on the extracted features, facilitates comparative evaluation of the proposed methods. In addition, Frequency Modulation features are combined in a complementary manner with the Mel Frequency CepstralCoefficients (MFCCs) to improve the performance of the ASV system under channel variability of various types. The proposed fused system shows a relative reduction of up to 23% in Equal Error Rate (EER) over the MFCC-based system when evaluated on the NIST 2008 dataset. Currently, the main challenge in speaker modelling is channel variability across different sessions. A recent approach to channel compensation, based on Support Vector Machines (SVM) is Nuisance Attribute Projection (NAP). The proposed multi-component approach to NAP, attempts to compensate for the main sources of inter-session variations through an additional optimization criteria, to allow more accurate estimates of the most dominant channel artefacts and to improve the system performance under mismatched training and test conditions. Another major issue in speaker recognition is that the variability of score distributions due to incompletely modelled regions of the feature space can produce segments of the test speech that are poorly matched to the claimed speaker model. A segment selection technique in score normalization is proposed that relies only on discriminative and reliable segments of the test utterance to verify the speaker. This approach is particularly useful in noisy conditions where using speech activity detection is not reliable at the feature level. Another source of score variability comes from the fact that not all phonemes are equally discriminative. To address this, a new score re-weighting technique is applied to likelihood values based on the discriminative level of each Gaussian component, i.e. each particular region of the feature space. It is found that a limited number of Gaussian mixtures, herein termed discriminative components are responsible for the overall performance, and that inclusion of the other non-discriminative components may only degrade the system performance.