Extracting the Electric Network Frequency (ENF) fluctuations from an audio recording and comparing it to a reference database is a new technology intended to perform forensic digital audio authentication. The objective of this thesis is to implement and design a range of programs and algorithms for capturing and extracting ENF signals. The developed C-program combined with a probe can be used to build the reference database. Our implementation of the Short-Time Fourier Transform method is intended for the ENF extraction of longer signals while our novel proposed use of the Autoregressive parametric method and our implementation of the zero-crossing approach tackle the case of shorter recordings. A Graphical User Interface (GUI) was developed to facilitate the process of extracting the ENF fluctuations. The whole process is tested and evaluated for various scenarios ranging from long to short recordings.

Cílem této práce byl vývoj programu pro automatickou detekci arousalu v signálu spánkového EEG s použitím metod časově-frekvenční analýzy. Předmětem studie bylo 13 celonočních polysomnografických nahrávek (čtyři svody EEG, EMG, EKG a EOG), tj. celkově více než 100 hodin záznamu. Jednalo se o část dat z dřívějších výzkumných prací expertní lékařky v problematice spánku Dr. Emilie Sforzy, Ženeva, Švýcarsko, která rovněž poskytla základní hodnocení těchto dat. V záznamech bylo celkem označeno 1551 arousal událostí. Pro usnadnění výběru konkrétní metody časově-frekvenční analýzy byla následně vytvořena sada nástrojů pro vizualizaci jednotlivých signálů a jejich různých časově-frekvenčních vyjádření. S ohledem na závěry vizuální analýzy, charakter signálu EEG a efektivitu výpočetních metod byla pro analýzu vybrána waveletová transformace s mateřskou vlnkou Daubechies řádu 6. Jednotlivé svody EEG byly dekomponovány do šesti frekvenčních pásem. Z takto odvozených signálů a signálu EMG byly následně stanoveny ukazatele možné přítomnosti události arousalu. Tyto ukazatele byly dále váhovány lineárním klasifikátorem, jehož hodnoty vah byly optimalizovány pomocí genetického algoritmu. Na základě hodnoty lineárního klasifikátoru bylo rozhodnuto o přítomnosti události arousalu v daném svodě EEG – arousal byl detekován, jestliže hodnota klasifikátoru překročila danou mez na dobu více než 3 a méně než 30 vteřin. V celém záznamu pak byl arousal označen, byl-li detekován alespoň v jednom ze svodů EEG. Následně byly odvozeny míry senzitivity a selektivity detekce, jež byly rovněž základem pro stanovení fitness funkce genetického algoritmu. Pro učení genetického algoritmu byly vybrány první čtyři záznamy. Na základě takto optimalizovaných vah vznikl program pro automatickou detekci, který na celém souboru 13 záznamů dosáhl ve srovnání s expertním hodnocením míry senzitivity 76,09%, selektivity 53,26% a specificity 97,66%.

Artificial Inteligens and machine learning has started to get established as reco- gnizable terms to the general masses in their daily lives. Applications such as voice recognicion and image recognicion are used widely in mobile phones and autonomous systems such as self-drivning cars. This study examines how one can utilize this technique to classify sound as a complement to videosurveillan- ce in different settings, for example a busstation or other areas that might need monitoring. To be able to do this a technique called Convolution Neural Ne- twork has been used since this is a popular architecture to use when it comes to image classification. In this model every sound has a visual representation in form of a spectogram that showes frequencies over time. One of the main goals of this study has been to be able to apply this technique on so called IOT units to be able to classify sounds in real time, this because of the fact that these units are relativly affordable and requires little resources. A Rasberry Pi was used to run a prototype version using tensorflow & keras as base api ́s. The studys re- sults show which parts that are important to consider to be able to get a good and reliable system, for example which hardware and software that is needed to get started. The results also shows what factors is important to be able to stream live sound and get reliable results, a classification models architecture is very important where different layers and parameters can have a large impact on the end result. / Termer som Artificiell Intelligens och maskininlärning har under de senaste åren börjat etablera sig hos den breda massan och är numera någonting som på- verkar nästan alla människors vardagliga liv i någon form. Vanliga använd- ningsområden är röststyrning och bildigenkänning som bland annat används i mobiltelefoner och autonoma system som självkörande bilar med mera. Den här studien utforskar hur man kan använda sig av denna teknik för att kunna klassi- ficera ljud som ett komplement till videoövervakning i olika miljöer, till exem- pel på en busstation eller andra övervakningsobjekt. För att göra detta har en teknik kallad Convolution Neural Network använts, vilket är en mycket populär arkitektur att använda vid klassificering av bilder. I denna modell har varje ljud fått en visuell representation i form av ett spektogram som visar frekvenser över tid. Ett av huvudmålen med denna studie har varit att kunna applicera denna teknik på så kallade IOT-enheter för att klassificera ljud i realtid. Dessa är rela- tivt billiga och resurssnåla enheter vilket gör dem till ett attraktivt alternativ för detta ändamål. I denna studie används en Raspberry Pi för att köra en prototyp- version med Tensorflow & Keras som grund APIer. Studien visar bland annat på vilka moment och delar som är viktiga att tänka på för att få igång ett smidigt och pålitligt system, till exempel vilken hårdvara och mjukvara som krävs för att starta. Den visar också på vilka faktorer som spelar in för att kunna streama ljud med bra resultat, detta då en klassifikationsmodells arkitektur och upp- byggnad kan ha stor påverkan på slutresultatet.

The primary goal of this work is to arrive at direction tomography (DT) algorithms freed from the severe linearization in the formulation, and as-assumptions on variation of the refractive index distribution (RID), involved in the earlier approaches based on Born and Royton approximations and the Fourier di reaction theorem (FDT). To start with, a direct single-step re-covery of RID from intensity measurements is demonstrated, replacing the common two-step procedure involving, rest the recovery of phase from in-density followed by the inversion of scattered led for the RID. The information loss, unavoidable in a two-step procedure is thus successfully addressed. Secondly, an iterative method which works with a forward model obtained directly from the Helmholtz equation is developed. This forward model, though has simplifying assumptions, is more general and can accommodate larger variations in RID than that allowed in the previous linear models. The iterative procedure has an update step which uses a linearization of the forward model and a re-linearization step at the updated RID. The procedure which directly employs the measured intensities is used as part of a deterministic Gauss-Newton algorithm and a stochastic optimization algorithm which uses the ensemble Kalman lter to arrive at the recursive update. The stochastic method is found to be more noise-tolerant and efficient to take care of process model inaccuracies. The proof is seen in better reconstructions from experimental data for two example objects, namely, a graded-index optical bre and a photonic-crystal bre. It is further ob-served that the reconstructions from photonic crystal bre are blurred, noisy and less accurate. Identifying the inaccurate implementation of the modemed Helmholtz equation for large k values employing the current sampling rate as the shortcoming, a new procedure, which splits the bandwidth into smaller components using short-time Fourier Transform is developed. The set of equations arrived at, each t for a narrow frequency band, is solved and the solutions are reassembled to obtain the scattered led for the original problem. The simulated di rated intensities so obtained are better matched to their measured experimental counterparts. However, the impel-mentation of the mode end procedure is computation-intensive, for which a parallel-processing machine can be a good solution. The recovery of RID with this mode cation is not attempted in this work and is left for future implementation.

Tato práce se zabývá vylepšením algoritmu pro sdružování (clustering) ERP signálů pomocí analýzy časových a prostorových vlastností pseudo-signálů získaných za pomocí metody analýzy nezávislých komponent (Independent Component Analysis). Naším zájmem je nalezení nových vlastností, které by zlepšily stávající výsledky. Tato práce se zabývá použitím Fourierovy transformace (Fourier Transform), FIR filtru a krátkodobé Fourierovy transformace ke zkvalitnění informace pro sdružovací algoritmy. Princip a použitelnost metody jsou popsány a demonstrovány ukázkovým algoritmem. Výsledky ukázaly, že pomocí dané metody je možné získat ze vstupních dat zajímavé informace, které mohou být úspěšně použity ke zlepšení výsledků.

Muscle fatigue is a severe problem for elite athletes, and this is due to the long resting times, which can vary. Various mechanisms can cause muscle fatigue which signifies that the specific muscle has reached its maximum force and cannot continue the task. This thesis was about surveying and exploring state-of-the-art methods and systematically, theoretically, and practically testing the applicability and performance of more recent machine learning methods on an existing EMG to muscle fatigue pipeline. Several challenges within the EMG domain exist, such as inadequate data, finding the most suitable model, and how they should be addressed to achieve reliable prediction. This required approaches for addressing these problems by combining and comparing various state-of-the-art methodologies, such as data augmentation techniques for upsampling, spectrogram methods for signal processing, and transfer learning to gain a reliable prediction by various pre-trained CNN models. The approach during this study was to conduct seven experiments consisting of a classification task that aims to predict muscle fatigue in various stages. These stages are divided into 7 classes from 0-6, and higher classes represent a fatigued muscle. In the tabular part of the experiments, the Decision Tree, Random Forest, and Support Vector Machine (SVM) were trained, and the accuracy was determined. A similar approach was made for the spectrogram part, where the signals were converted to spectrogram images, and with a combination of traditional- and intelligent data augmentation techniques, such as noise and DCGAN, the limited dataset was increased. A comparison between the performance of AlexNet, VGG16, DenseNet, and InceptionV3 pre-trained CNN models was made to predict differences in jump heights. The result was evaluated by implementing baseline classifiers on tabular data and pre-trained CNN model classifiers for CWT and STFT spectrograms with and without data augmentation. The evaluation of various state-of-the-art methodologies for a classification problem showed that DenseNet and VGG16 gave a reliable accuracy of 89.8 % on intelligent data augmented CWT images. The intelligent data augmentation applied on CWT images allows the pre-trained CNN models to learn features that can generalize unseen data. Proving that the combination of state-of-the-art methods can be introduced and address the challenges within the EMG domain.

This thesis focuses on the simulation of the gyroscopic effect using the software MSC Adams. A simple shaft-disk system was created and parameter of the sys-tem were changed in order to study the influence of the gyroscopic effect. It was shown that an increasing bearing stiffness reduces the precession motion. Fur-thermore, it was shown that the gyroscopic effect vanishes if the disk of system is placed symmetrically on the shaft, which reduces the system to a Jeffcott-Ro-tor. The second objective of this study was to analyze different defects in a simple fixed axis gear set. In particular, a cracked shaft, a cracked pinion and a chipped pinion as well as a healthy gear system were created and tested in Adams. The contact force between the two gears was monitored and the 2D and 3D frequency spectrum, as well as the Wavelet Transform, were plotted in order to compare the individual defects. It was shown that the Wavelet Transform is a powerful tool, capable of identifying a cracked gear with a non-constant speed. The last part of this study included fault detection with statistical methods as well as with the Sideband Energy Ratio (SER). The time domain signal of the individual faults were used to compare the mean, the standard deviation and the root mean square. Furthermore, the noise profile in the frequency spectrum was tracked with statistical methods using the mean and the standard deviation. It was demonstrated that it is possible to identify a cracked gear, as well as a chipped gear, with statistical methods. However, a cracked shaft could not be identified. The results also show that SER was only capable to identify major defects in a gear system such as a chipped tooth.

This thesis describe detection of direction receiving sound with small–size microphone array. Work is based on analyzing methods of time delay estimation, energy decay or phase difference signal. Work focus mainly on finding of angle of arrival in small time difference. Results of measuring, as programming sound, so sound recorded in laboratory conditions and real enviroment, are mentioned in conclusion. All calculations were done by platform Matlab

Master‘s thesis reviews the characteristics of the AVR32 architecture, AVR32UC microarchitecture, and especially AT32UC3A0512 microcontroller. This microcontroller is mounted on the board EVK1100, which is used for debugging applications. The entire analysis is focused on the ability to process audio signals on this board. For the board is created AD/DA interface and its control library. Follows necessary description of used DSP-lib library. The last part is a description of the theory and implementation of two sound effects and implementation of operating system FreeRTOS.

Dans ce travail, une approche de reconnaissance de l’orateur en utilisant un microphone de contact est développée et présentée. L'élément passif de contact est construit à partir d'un matériau piézoélectrique. La position du transducteur piézoélectrique sur le cou de l'individu peut affecter grandement la qualité du signal recueilli et par conséquent les informations qui en sont extraites. Ainsi, le milieu multicouche dans lequel les vibrations des cordes vocales se propagent avant d'être détectées par le transducteur est modélisé. Le meilleur emplacement sur le cou de l’individu pour attacher un élément transducteur particulier est déterminé en mettant en œuvre des techniques de simulation Monte Carlo et, par conséquent, les résultats de la simulation sont vérifiés en utilisant des expériences réelles. La reconnaissance est basée sur le signal généré par les vibrations des cordes vocales lorsqu'un individu parle et non sur le signal vocal à la sortie des lèvres qui est influencé par les résonances dans le conduit vocal. Par conséquent, en raison de la nature variable du signal recueilli, l'analyse a été effectuée en appliquant la technique de transformation de Fourier à court terme pour décomposer le signal en ses composantes de fréquence. Ces fréquences représentent les vibrations des cordes vocales (50-1000 Hz). Les caractéristiques en termes d'intervalle de fréquences sont extraites du spectrogramme résultant. Ensuite, un vecteur 1-D est formé à des fins d'identification. L'identification de l’orateur est effectuée en utilisant deux critères d'évaluation qui sont la mesure de la similarité de corrélation et l'analyse en composantes principales (ACP) en conjonction avec la distance euclidienne. Les résultats montrent qu'un pourcentage élevé de reconnaissance est atteint et que la performance est bien meilleure que de nombreuses techniques existantes dans la littérature. / In this work, a speaker recognition approach using a contact microphone is developed and presented. The contact passive element is constructed from a piezoelectric material. In this context, the position of the piezoelectric transducer on the individual’s neck may greatly affect the quality of the collected signal and consequently the information extracted from it. Thus, the multilayered medium in which the sound propagates before being detected by the transducer is modeled. The best location on the individual’ neck to place a particular transducer element is determined by implementing Monte Carlo simulation techniques and consequently, the simulation results are verified using real experiments. The recognition is based on the signal generated from the vocal cords’ vibrations when an individual is speaking and not on the vocal signal at the output of the lips that is influenced by the resonances in the vocal tract. Therefore, due to the varying nature of the collected signal, the analysis was performed by applying the Short Term Fourier Transform technique to decompose the signal into its frequency components. These frequencies represent the vocal folds’ vibrations (50-1000 Hz). The features in terms of frequencies’ interval are extracted from the resulting spectrogram. Then, a 1-D vector is formed for identification purposes. The identification of the speaker is performed using two evaluation criteria, namely, the correlation similarity measure and the Principal Component Analysis (PCA) in conjunction with the Euclidean distance. The results show that a high percentage of recognition is achieved and the performance is much better than many existing techniques in the literature.