Thesis (doctoral)--Rijksuniversiteit Groningen, 1997. / Includes bibliographical references (p. [393]-428) and index.

The last few years have seen the start of a unique change in microphones for consumer devices such as smartphones or tablets. Almost all analogue capacitive microphones are being replaced by digital silicon microphones or MEMS microphones. MEMS microphones perform differently to conventional analogue microphones. Their greatest disadvantage is significantly increased self-noise or decreased SNR, while their most significant benefits are ease of design and manufacturing and improved sensitivity matching. This thesis presents research on speech processing, comparing conventional analogue microphones with the newly available digital MEMS microphones. Specifically, voice activity detection, speaker diarisation (who spoke when), speech separation and speech recognition are looked at in detail. In order to carry out this research different microphone arrays were built using digital MEMS microphones and corpora were recorded to test existing algorithms and devise new ones. Some corpora that were created for the purpose of this research will be released to the public in 2013. It was found that the most commonly used VAD algorithm in current state-of-theart diarisation systems is not the best-performing one, i.e. MLP-based voice activity detection consistently outperforms the more frequently used GMM-HMM-based VAD schemes. In addition, an algorithm was derived that can determine the number of active speakers in a meeting recording given audio data from a microphone array of known geometry, leading to improved diarisation results. Finally, speech separation experiments were carried out using different post-filtering algorithms, matching or exceeding current state-of-the art results. The performance of the algorithms and methods presented in this thesis was verified by comparing their output using speech recognition tools and simple MLLR adaptation and the results are presented as word error rates, an easily comprehensible scale. To summarise, using speech recognition and speech separation experiments, this thesis demonstrates that the significantly reduced SNR of the MEMS microphone can be compensated for with well established adaptation techniques such as MLLR. MEMS microphones do not affect voice activity detection and speaker diarisation performance.

<p>This report describes how wavefiles can be recorded with programs written in the C++ language . The report also describes how the</p><p>recorded sound is handled before it can be analyzed. The report describes the different parts that make a wave file to a wave file and</p><p>how Windows handle these files. There is always some silence or noise of unknown length that passes through the system before</p><p>some interesting sound arrives. That silence or nose must not be recorded and therefore some function must be developed that can</p><p>detect interesting sound or speech. Besides of that this report handle the problem of cutting a longer wave file into smaller parts and</p><p>how to down sample recorded files and files that have been cut.</p>

Barn med någon form av språkstörning är förekommande i förskolan. Dock är det inte förrän när barnet är runt 4-5 års ålder som dessa språkstörningar brukar upptäckas. Då de kan vara svåra att upptäcka innan eftersom det är svårt att säga om det rör sig om en störning eller om barnet inte är färdigutvecklat. Därför är det viktigt att förskollärare tänker på hur de stimulerar barns språkutveckling för att barnen skall kunna utveckla sitt språk. Syftet med denna studie är att undersöka om arbetet med barn som har språkstörningar inom förskolan bedrivs så som forskningen idag rekommenderar eller om det kunde förbättras. Resultatet visar att förskolan ofta arbetar så som forskningen anser att man bör göra, men att det finns punkter som förskolorna kan förbättra för att barn med någon form av språkstörning ska utveckla sitt språk.

This master thesis deals with methods of automatic recognition of bird species by their voices. In first, I defined the database of records and created a reference data by handmade evaluation. The next step is to find the optimal features for describing a bird singing. I use a Human Frequency cepstral Coefficients (HFCC). For the best accuracy of recognition is necessary to correctly classify a bird's vocalization from a non-vocalization segments. The VAD system is based on an algorithm k-Nearest Neighbours. The last step describes the system based on Hidden Markov Models which allows to recognize the concrete bird species from the parts of bird's singing.

This thesis deals with the problematics of the use of the FMEA method in the manufacturing process of engineering company, which is a certified supplier for the automotive industry. The first part of the thesis is focused on the description of the method , its division and its application. The following part of the thesis contains a description of some basic quality tools, which are used a tool for ist preparation. In the practical part of the thesis there is a decription of the application of the FMEA method process on the specific product by Metaldyne spol. s r.o. The collected data are used for the preparation of the FMEA form, in which precautions to reduce nonquality in the manufacturing proces are determined. The final part of the thesis deals with the technical and economic evaluation of profuction of the lower body of a hydraulic pressure filter PAL2R.

The VAD Wind Profile (VWP), a time-height display of winds computed by the National Weather Service's WSR-88D radar, is known on occasion to have errors at night during the fall and spring seasons. Several studies, such Haro and Gauthreaux (1998), confirm that migrating birds often contaminate the VWP output. By means of telescopic observations of a full moon, birds were observed flying on two nights when VWP contamination was suspected. The nature of the VWP errors is consistent with migrating birds due to the seasonality, nocturnal nature, and the magnitude of the errors found (greater than 10 knots). With careful selection of data, two clusters of points on the Velocity-Azimuth Display (VAD) are found to exist at certain altitudes when birds begin migrating. One cluster of points is due to radar sample volumes containing birds, and the other cluster is from radar sample volumes without birds. Being able to determine which cluster represents the wind could allow the wind to be calculated by the VWP. Present limitations with the Radar Product Generator's processor and memory prohibit a very advanced detection algorithm. Two simple objective techniques to determine the existence of the two clusters, and determine the wind, were tested. While they show some promise, these methods require further operational testing to determine their usefulness for real-time warning of bird contamination and the reporting of the true wind.

Few therapeutic alternatives exist for patients with a failing single ventricle physiology. To address this need, this thesis project investigated two new therapeutic alternatives, which sought to positively augment the Fontan hemodynamics. The first modality introduced a non-invasive method of external pressure application to the lower extremities. A clinical study (n=2) was conducted, and results indicated an increase in flow as a consequence to an increase in transmural pressure in the lower extremities. The second modality investigated a minimally invasive blood pump. Numerical analyses of the pump were performed to examine hydraulic performance under physiologic conditions. The pump produced pressure rises of 1 to 25 mmHg over flows of 1 to 4 LPM, has a blood damage index less than 1% and was also found to successfully augment the hydraulic energy of the Fontan physiology. This work resulted in substantial progress to develop both modalities and address a significant human health problem.

O avanço da bioengenharia atual tem sido motivado pela crescente necessidade da humanidade em se buscar formas para amenizar o sofrimento, garantir tempo para um tratamento, melhorar a qualidade de vida ou sanar quadros clínicos de pacientes. Assim, a crescente necessidade de órgãos artificiais impulsiona a área de bioengenharia para que seja dado um suporte no desenvolvimento destes mecanismos. Neste contexto se encontra o presente trabalho. Aqui, apresenta-se uma metodologia numérica para que seja aplicada a desenvolvimento e testes de Dispositivos de Assistência Cardíaca, de forma a otimizar o processo de desenvolvimento e estimar os possíveis problemas inerentes ao seu funcionamento. Compondo a metodologia, uma ampla discussão de cada etapa numérica foi elaborada, contribuindo para uma metodologia flexível para uma ampla variedade de aplicações. A hemólise também foi investigada através dos principais modelos da literatura, bem como foram propostos modelos adaptados para tentar estimar a hemólise verificada experimentalmente. Além das metodologias para cada etapa, uma metodologia geral utilizando Sistemas de Referências Múltiplas (SRM) compõe os resultados, para uma sequência de passos numéricos que possa obter resultados satisfatórios em comparação com resultados de bancada por loop test. Outros resultados encontrados e devidamente discutidos foram inerentes a uma densidade de malha a ser utilizada para que ocorra simulações com independência de malha, na qual uma densidade a partir de 318,43 elementos/mm³ para referenciais não-inerciais é proposta. O mapeamento da turbulência por seis modelos que aplicam a média de Reynolds também é discutido, indicando os melhores modelos a serem empregados para um número de Reynolds relativo (Re*) que relaciona a influência dos contornos nos resultados numéricos obtidos. Além disso, uma nova abordagem por tensão fisiológica é proposta para o cálculo da hemólise, sendo comparada aos modelos clássicos adotados. Os resultados para hemólise indicam um bom ajuste para o novo modelo proposto, bem como indica o correto tratamento de unidades para os modelos tradicionais. Por fim, as análises recaem na melhoria do dispositivo e conclui a metodologia numérica a ser empregada, preenchendo lacunas em cada etapa e determinando uma maneira de análise numérica cujos resultados são confiáveis. / The advance of current bioengineering has been driven by the increasing need of humanity to seek ways to alleviate the suffering, ensure time to treatment, improve the quality of life or cure medical conditions of patients. Thus, the growing need for artificial organs boosts bioengineering area to be given a support in the development of these mechanisms. In this context is the present work. Here, we present a numerical methodology to be applied to development and Cardiac Assist Devices tests in order to optimize the development process and estimate the potential problems inherent in their operation. Compounding the methodology, a comprehensive discussion of each numerical step was developed, contributing to a flexible methodology for a wide variety of applications. Hemolysis was also investigated through the main models of literature, and have been proposed models adapted to try to estimate hemolysis verified experimentally. In addition to the methodologies for each step, a general methodology using Multiple Reference Systems (MRF) makes up the results for a sequence of numeric steps you can get satisfactory results compared to bench test results for loop. Other findings were discussed and properly attached to a mesh density being used for simulations occurring independently mesh in which a density from 318.43 elements/mm³ Non-inertial frames is proposed. The mapping of turbulence six models applying Reynolds medium is also discussed, indicating the best designs to be employed for a number of relative Reynolds (Re*) that relates the influence of the contours of numerical results obtained. In addition, a new approach by physiological stress is proposed for the calculation of hemolysis, being compared to classic models adopted. The results for hemolysis indicate agreement with the proposed new model, and indicates the correct treatment units to the traditional models. The final analyses fall into improved device and completes the numerical methodology to be used by filling gaps in each step and determining a way to numerical analysis results which are dependable.

<p>Detta examensarbete har vägletts av följande syfte och frågeställning: Genom att undersöka hur lärare i tre F-6 skolor uppfattar arbetsmaterialet MultiMatte, syftar studien att söka spegla lärares uppfattningar om arbetsmaterialet ”MultiMatte” ur ett didaktiskt perspektiv.</p><p>Kvalitativa intervjuer har sedan genomförts med sex stycken lärare. Inspirerade av den reflexiva intervjun har vi analyserat och tolkat lärarnas uppfattningar om MultiMatte och på så vis uppmärksammat deras åsikter. Lärarnas uppfattningar har sedan analyserats, först med hjälp av den lodräta analysen och därefter den vågräta analysen. Svaren har sedan kategoriserats in i tre områden som beskrivits och rubricerats som: utmärkande drag hos MultiMatte enligt de intervjuade lärarna, vilka följder får det för lärare som använder sig av MultiMatte i sitt arbete och vilka följder får det för elever som arbetar med MultiMatte. Det kom bland annat fram i intervjuundersökningen att lärarna anser att MultiMatte tränar eleverna i att tänka, prata och förklara matematik. Det framkommer även att det är svårt för vikarier att hoppa in i klassen för de är inte insatta i MultiMattes sätt att fungera. I diskussionen har slutligen resultatet från undersökningen relaterats till vår bakgrund. Vi har använt oss av de didaktiska frågorna vad, hur och varför som ett verktyg, för att tydliggöra och organisera vår diskussion, i den framträder även reflektioner från oss. Vi har bland annat reflekterat över att något som kan påverka detta material tämligen snart är de organisatoriska ramar som finns i skolan. Genom att vi att problematiserat de didaktiska frågorna i relation till lärarnas uppfattningar och arbetsmaterialet MultiMatte menar vi, att det går att motivera användandet av MultiMatte i sin undervisning, men att ramarna i skolan inte alltid tillåter det.</p>