Dissertations / Theses on the topic 'Reverberation'

We conduct a multiwavelength continuum variability study of the Seyfert 1 galaxy NGC 5548 to investigate the temperature structure of its accretion disk. The 19 overlapping continuum light curves (1158 angstrom to 9157 angstrom) combine simultaneous Hubble Space Telescope, Swift, and ground-based observations over a 180 day period from 2014 January to July. Light-curve variability is interpreted as the reverberation response of the accretion disk to irradiation by a central time-varying point source. Our model yields the disk inclination i = 36 degrees +/- 10 degrees, temperature T-1= (44 +/- 6) x 10(3) K at 1 light day from the black hole, and a temperature-radius slope (T proportional to r(-alpha)) of alpha = 0.99 +/- 0.03. We also infer the driving light curve and find that it correlates poorly with both the hard and soft X-ray light curves, suggesting that the X-rays alone may not drive the ultraviolet and optical variability over the observing period. We also decompose the light curves into bright, faint, and mean accretion-disk spectra. These spectra lie below that expected for a standard blackbody accretion disk accreting at L/L-Edd = 0.1.

The reverberation-room method, which assumes a diffuse sound field, has long been used for various standardized room-acoustical determinations – e.g. of surface-absorption coefficients, power levels of sound sources, transmission losses of acoustical partitions, etc. In this regard, a number of standards have emerged to offer some help by outlining necessary reverberation-room design guidelines to achieve sufficient sound-field diffuseness. However, unsatisfactory opinions regarding the prediction accuracy of the method, especially at low frequencies, have been reported over the years. This might be due to deviations from the assumed diffuse-field concept, which is very challenging to implement from an application point of view; also there are no straight-forward ways to characterize the degree of sound-field diffuseness. To investigate the problem and propose solutions, diffuse-field theory and existing standards have been revisited. Using numerical, finite-element-based, modal prediction, their capacity/effectiveness to achieve a diffuse sound field is analyzed by means of a number of descriptors (room-acoustical parameters). Because of time limitations, the concept regarding the design of a reverberation-room structure – i.e. size, shape, etc. – is mainly explored, rather than the internal test-setup arrangements. The prediction accuracy of different room-acoustical parameters are also determined by the reverberation-room standard methods, with respect to both the Sabine and Eyring versions of the diffuse-field formulae, and both in octave and third-octave bands. The minimum approachable frequencies of predictions and the quality of sound-field diffuseness are discussed in terms of the prediction accuracy of different room-acoustical parameters. Considering three room volumes prescribed by standards, and four room shapes for each of the volumes, it has been found that the reverberation room of volume 150 m³, as prescribed by the ISO 354 standard, with the typical dimensional orientation (longest x-dimension/shortest vertical dimension) yields better field diffuseness than the other rooms of different volumes and shapes. To check the possibility of further improvement of the field diffuseness, a number of additional features are integrated into that reverberation room. It is found that the rooms with diffusers and absorbent corner treatments yield improved sound-field diffuseness, while the rooms with diffuse surface reflection yield poor field diffuseness due to the increased surface absorption.
Applied Science, Faculty of
Mechanical Engineering, Department of
Graduate

Reverberation enhancement is a technology which allows the reverberation time of a room to be increased through the use of an electronic system. These systems have traditionally been applied to improve the acoustics of large concert halls but the technology can also be used in smaller spaces with several possible applications. Previous uses of reverberation enhancement in small rooms have largely consisted of direct transplants of systems designed for large concert halls. This work investigates the complications which arise when using reverberation enhancement in a small room due to the differences in the acoustic properties of the space and also the restriction on the channel count of the system due to physical constraints. The first part of this work deals with increasing the resultant reverberation time of the room without requiring additional system channels. This is achieved through the use of processing within the system. Two methods have been investigated. The first extends the resultant reverberation time without changing the feedback gain. The processing used for this purpose is either electronic reverberation or simple delay, both of which have been shown to allow significant increases in resultant reverberation time. These changes can be predicted accurately using diffuse field theory. The other method uses time-varying processing to increase the maximum stable feedback gain. This has been shown to allow increases in resultant reverberation time but also causes undesirable artefacts which limit the usability of this technique. The second part of this work focuses on the differences in the acoustic properties of small rooms and especially the ways in which these rooms differ from a diffuse field. This includes the consideration of the modal properties of the room at low frequency which are insignificant in a large room. It has been shown that the spatial and frequency variations of the room at low frequency can be reduced through numerical optimisation of the processing within the reverberation enhancement system. Finally, the diffusion of the sound field and the early energy in the impulse response have been considered. It is shown that restrictions on the resultant reverberation time may be required in order to create a subjectively acceptable acoustic response. Overall, this work has shown that by accounting for the properties of the room, excellent performance of the system can be achieved.

The use of reverberation models is required in many applications such as acoustic measurements,speech dereverberation and robust automatic speech recognition. The aim of this thesis is toinvestigate different models and propose a perceptually-relevant reverberation model with suitableparameter estimation techniques for different applications. Reverberation can be modelled in both the time and frequency domain. The model parametersgive direct information of both physical and perceptual characteristics. These characteristicscreate a multidimensional parameter space of reverberation, which can be to a large extent capturedby a time-frequency domain model. In this thesis, the relationship between physical and perceptualmodel parameters will be discussed. In the first application, an intrusive technique is proposed tomeasure the reverberation or reverberance, perception of reverberation and the colouration. Theroom decay rate parameter is of particular interest. In practical applications, a blind estimate of the decay rate of acoustic energy in a roomis required. A statistical model for the distribution of the decay rate of the reverberant signalnamed the eagleMax distribution is proposed. The eagleMax distribution describes the reverberantspeech decay rates as a random variable that is the maximum of the room decay rates and anechoicspeech decay rates. Three methods were developed to estimate the mean room decay rate fromthe eagleMax distributions alone. The estimated room decay rates form a reverberation model thatwill be discussed in the context of room acoustic measurements, speech dereverberation and robustautomatic speech recognition individually.

Room reverberation adds slowly-decaying 'tails' at the end of sounds, yet listeners do not perceive these strongly. This is evinced by loudness judgements of stimuli shaped with 'reverberant tail-like' slow offsets being quieter than their reversed counterparts. This study investigates whether a perceptual 'constancy' mechanism is responsible, accounting for reverberation effects and maintaining constant perception despite changes to the sound's physical characteristics. Such a mechanism might take account of reverberation by separating sounds with decaying rails into source characteristics and effects from reverberation, then dismissing energy within tails from listeners' loudness judgements. The process appears to be informed by preceding contexts because loudness differences are enhanced following 'standard' stimuli with similar tail• like decays. This study found that this 'loudness context effect' is more pronounced when using stimuli with real• reverberant tails than artificially-shaped offsets, indicating the effect is sensitive to more than simplified slow offsets. Additionally, reverberation decorrelates sounds at a listener's two ears. When binaural and monaural conditions are examined, a 'binaural de-reverberation' occurs whereby perception of uncorrelated binaural tails is further reduced, causing a decrease in the loudness context effect. However, the loudness context effect is only apparent in situations where both standard and test stimuli occupy the same narrow frequency band and is markedly reduced in widely separated cross-band conditions. These monaural and within• band loudness effects are similar to a constancy for reverberation found in speech perception in which reverberant tails are also dismissed. The underlying processes causing these effects are explored. While models of the early auditory system can account for some of these loudness differences, such models are insensitive to additional influence from contexts. It seems a further, higher-level process may be having an effect: one that uses information from previously heard stimuli, such as the presence of similarly tailed decays, binaural comparisons, and stimuli sharing a common frequency

Thesis (M.S. in Electrical Engineering and M.S. in Engineering Acoustics) Naval Postgraduate School, June 1994.
Thesis advisor(s): James H. Miller, C. Chiu. "June 1994." Bibliography: p. 69-71. Also available online.

The rapid expansion of electronic industry calls for effective and efficient electromagnetic (EM) measurements, including the characterization of devices under test (DUT), such as antennas or wireless devices, and the electromagnetic compatibility (EMC) testing. In the real world, EM measurements can be influenced by a number of uncontrollable factors which will afflict the measurements. These factors make the measurements very difficult especially when the measurements require high precision and/or low power relative to the background noise. To conduct EM measurements accurately, many different facilities/environments have been developed, including anechoic chambers (ACs), transverse electromagnetic (TEM) Cells, and reverberation chambers (RCs). These three environments have different characteristics. Over the past several decades, RCs have been enjoying growing popularity as a promising facility for the characterization of wireless devices and for the EMC testing. The RC measurement method exhibits much competitive superiority over the AC method and TEM Cell method, such as low cost, enhanced test repeatability, a more realistic test environment, and easily achieved high-field environment. The application of the RC for performing EMC testing was first proposed by H. A. Mendes in 1968. In the recent IEC 61000-4-21 standard, the importance of EMC testing using RCs as an alternative measurement technique has been recognized. To make the RC well stirred, a large number of independent samples (stirrer positions) are required. Consequently, the measurement time is usually long (typically several hours), which has greatly restricted the engineering applications of the RC measurement techniques. The purpose of this thesis is to present our studies on improving the measurement efficiency of RCs in recent years, including the efficient measurement of the averaged absorption cross section (ACS) with only one antenna, the rapid volume measurement method using the averaged ACS, the simplified shielding effectiveness (SE) measurement using the nested RC with two antennas, and the improved antenna array efficiency measurement in an RC. For ACS measurement, the proposed one-antenna methods in both the frequency domain and the time domain are presented. The measurement setup is greatly simplified and the measurement time is significantly shortened. The efficient measurement of the ACS can be used to obtain the volume of a chamber, which leads to the rapid volume measurement method. For the SE measurement of electrically large enclosures using a nested RC, four improved measurement methods are proposed. Both the frequency-domain and time-domain methods are studied. The proposed methods require only two antennas and provide efficient measurement of SE without losing the accuracy. Finally, the accurate array efficiency measurement method in an RC using a power divider is presented. A power divider is used to excite the feeding ports of the array elements simultaneously. Thus, the efficiency measurement of the entire array can be effectively treated in a manner similar to a single port antenna, which would simplify the measurement procedure and reduce the overall measurement time. By introducing proper attenuators between the array elements and the power divider to alleviate the effect of the reflected power from the array to the insertion loss of the power divider, the array efficiency can be measured accurately even when the elements of the array are not well-matched with the power divider. The proposed method is advantageous especially for wideband antenna arrays where good impedance matching of array elements is difficult to maintain. In this thesis, it is shown that our proposed methods have greatly improved the RC measurement efficiency and simplified the measurement setup at the same time. These contributions could promote the industrial application of RCs.

The aim of this thesis was to make a practical tool for low frequency analysis in room acoustics.The need arises from Acad’s experience that their results from simulations using raytracing software deviate in the lower frequencies when compared to field measurements inrooms. The tool was programmed in Matlab and utilizes the Finite Difference Time Domain (FDTD) method, which is a form of rapid finite element analysis in the time domain.A number of tests have been made to investigate the practical limitations of the FDTD method, such as numerical errors caused by sound sources, discretization and simulation time. Boundary conditions, with and without frequency dependence, have been analysed bycomparing results from simulations of a virtual impedance tube and reverberation room to analytical solutions. These tests show that the use of the FDTD method appears well suited for the purpose of the tool.A field test was made to verify that the tool enables easy and relatively quick simulations of real rooms, with results well in line with measured acoustic parameters. Comparisons of the results from using the FDTD method, ray-tracing and finite elements (FEM) showed goodcorrelation. This indicates that the deviations Acad experience between simulated results and field measurements are most likely caused by uncertainties in the sound absorption data used for low frequencies rather than by limitations in the ray-tracing software. The FDTDtool might still come in handy for more complex models, where edge diffraction is a more important factor, or simply as a means for a “second opinion” to ray-tracing - in general FEM is too time consuming a method to be used on a daily basis.Auxiliary tools made for importing models, providing output data in the of room acoustic parameters, graphs and audio files are not covered in detail here, as these lay outside the scope of this thesis.
Målet för detta examensarbete var att undersöka möjligheten att programmera ett praktisktanvändbart verktyg för lågfrekvensanalys inom rumsakustik. Behovet uppstår från Acadserfarenhet att resultat från simuleringar med hjälp av strålgångsmjukvara avviker i lågfrekvensområdeti jämförelse med fältmätningar i färdigställda rum. Verktyget är programmerati Matlab och använder Finite Difference Time Domain (FDTD) metoden, vilket är en typav snabb finita elementanalys i tidsdomänen.En rad tester har genomförts för att se metodens praktiska begräsningar orsakade av numeriskafel vid val av ljudkälla, diskretisering och simuleringstid. Randvillkor, med och utanfrekvensberoende, har analyserats genom jämförelser av simulerade resultat i virtuella impedansröroch efterklangsrum mot analytiska beräkningar. Testerna visar att FDTD-metodentycks fungerar väl för verktygets tilltänkta användningsområde.Ett fälttest genomfördes för att verifiera att det med verktyget är möjligt att enkelt och relativtsnabbt simulera resultat som väl matcher uppmätta rumsakustiska parametrar. Jämförelsermellan FDTD-metoden och resultat beräknade med strålgångsanalys och finita elementmetoden(FEM) visade även på god korrelation. Detta indikerar att de avvikelser Acaderfar mellan simulerade resultat och fältmätningar troligen orsakas av osäkerheter i den ingåendeljudabsorptionsdata som används för låga frekvenser, snarare än av begränsningar istrålgångsmjukvaran. Verktyget kan fortfarande komma till användning för mer komplexamodeller, där kantdiffraktion är en viktigare faktor, eller helt enkelt som ett sätt att få ett”andra utlåtande” till resultaten från strålgångsmjukvaran då FEM-analys generellt är en förtidskrävande metod för att användas på daglig basis.Kringverktyg skapade för t.ex. import av modeller, utdata i form av rumsakustiska parametrar,grafer och ljudfiler redovisas inte i detalj i denna rapport eftersom dessa ligger utanförexamensarbetet.

In Yoshii et al., we described a new method for measuring extragalactic distances based on dust reverberation in active galactic nuclei (AGNs), and we validated our new method with Cepheid variable stars. In this Letter, we validate our new method with Type Ia supernovae (SNe Ia) that occurred in two of the AGN host galaxies during our AGN monitoring program: SN 2004bd in NGC 3786 and SN 2008ec in NGC 7469. Their multicolor light curves were observed and analyzed using two widely accepted methods for measuring SN distances, and the distance moduli derived are m= 33.47 +/- 0.15 for SN 2004bd and 33.83 +/- 0.07 for SN 2008ec. These results are used to obtain independently the distance measurement calibration factor, g. The g value obtained from the SN Ia discussed in this Letter is gSN= 10.61 +/- 0.50, which matches, within the range of 1s uncertainty, gDUST = 10.60, previously calculated ab initio in Yoshii et al. Having validated our new method for measuring extragalactic distances, we use our new method to calibrate reverberation distances derived from variations of Ha emission in the AGN broad-line region, extending the Hubble diagram to z approximate to 0.3 where distinguishing between cosmologies is becoming possible.

Halmstad University is currently equipped with an Echo-free chamber to perform EMC testing, but no reverberation chamber. The construction of a paddle stirrer to be utilized in Halmstad University would drastically reduce the time required to perform EMC testing, since reverberation chambers are more efficient than Echo-free chambers for these kind of tests. The goal of this project was to design and develop a paddle stirrer structure and a control system able to rotate the stirrer to specific repeatable absolute angles with an accuracy of a tenth of a degree,holding a mass up to 70kg and rotating a mass up to 20 kg distributed on a 1x1m surface. To achieve this goal the system has been designed using a metal base structure, bearings to hold the lower shaft in its axes reducing its friction, a stepper motor connected to the gears to increase its holding torque, a magnetic rotary encoder and a control system with a double feedback from interrupts and from the encoder to improve the accuracy and reliability of the system. The resultis a completely working prototype, which fulfils all the requirements except for the speed. The target speed has not been achieved due to the insufficient holding torque of the available stepper motor.
Halmstad Högskola är för närvarande utrustad med en Ekofri kammare för att utföra EMCtestning, men ingen modväxlande kammare. Konstruktionen av en paddle-omrörare för användning i Halmstad Högskola skulle drastiskt minska tiden som krävs för att utföra EMC-testning, eftersom modväxlande kammaren är effektivare än en ekofri kammare för dessa typer av tester.Målet med detta projekt var att designa och utveckla en paddle-omrörare, både strukturen samt ett styrsystem som kan rotera den till specifika repeterbara absoluta vinklar med en noggrannhet av en tiondel av en grad, hantera en massa upp till 70 kg och rotera en massa upp till 20 kg fördelad på en 1x1m yta. För att uppnå detta mål har systemet konstruerats med en basstruktur gjord av metall, lager har placerats i axeln för att reducera friktion, en stegmotor är anslutentill kugghjul för att öka vridmomentet, en magnetisk roterande sensor och ett styrsystem med en dubbel återkoppling från interrupts och från sensorn för att förbättra systemets noggrannhet och tillförlitlighet. Resultatet är en helt fungerande prototyp som uppfyller alla krav, förutom hastigheten. Målhastigheten har inte uppnåtts på grund av otillräckligt vridmoment hos stegmotorn.

Thesis (M.S. in Engineering Acoustics) Naval Postgraduate School, December 1991.
Thesis advisor(s) Coppens, Alan B. ; Sanders, James V. "December 1991." Includes bibliographical referenes. Also available online.

Two different chambers are studied in this thesis: the anechoic chamber (AC) and the reverberation chamber (RC). The AC has been developed for many years, while in the past few years RC has emerged as a promising facility not just in electromagnetic compatibility (EMC) measurements but also as a multi-disciplinary research facility in various areas. For the anechoic chamber, a CAD tool is developed to aid the design of an anechoic chamber. The objective is to estimate the chamber performance accurately. The ultimate goal is to minimise the cost but optimise the chamber performance for given conditions and specifications. This is very important for a commercial company. The CAD tool is developed based on the GO theory, two different algorithms are realised, acceleration techniques are applied, and measurements are performed to validate the CAD tool. For the reverberation chamber, a series of new measurement methods in the RC are developed including antenna radiation efficiency measurement, diversity gain measurement, radiated emission measurement, material characterisation, shielding effectiveness, volume measurement, etc. Finally, we apply the B-scan in an RC to characterise the behaviour of the electric field in the time domain. Statistical characterisation of the electric field in the time domain is given, stirrer efficiency is quantified based on the total scattering cross section (TSCS) of stirrers, and time gating technique in the RC is introduced. It has been found that the stirrer efficiency can be well-quantified in the time domain and the definition of stirrer efficiency in this thesis provides a universal and quantitative way to compare the performance between different RCs or different stirrer designs in one RC.

Narrow Line Type-1 Seyfert active galaxies can exhibit a high degree of variability in the X-ray regime. This thesis examines that variability in the context of reverberation models, in which a flare in activity has an extended, energy dependent, response. A novel method is developed for estimating the response function in different energy bands. This method is then applied to three AGN: Ark 564, 1H 0707{495, and NGC 4051. The striking evidence for reverberation revealed in Ark 564 leads to a more thorough examination of that object, combining spectral and temporal approaches to develop a plausible physical model for its behaviour. The preferred model is one in which the reverberation is due to scattering from hot Comptonizing material approximately 1500 light-seconds from the central source. This conclusion is reinforced by a simulation of the angular dependence of reflection by Comptonizing gas.

We present reverberation mapping results from the first year of combined spectroscopic and photometric observations of the Sloan Digital Sky Survey Reverberation Mapping Project. We successfully recover reverberation time delays between the g+i band emission and the broad H beta emission line for a total of 44 quasars, and for the broad Ha emission line in 18 quasars. Time delays are computed using the JAVELIN and CREAM software and the traditional interpolated cross-correlation function (ICCF): using well-defined criteria, we report measurements of 32 H beta and 13 Ha lags with JAVELIN, 42 H beta and 17 Ha lags with CREAM, and 16 H beta and eight Ha lags with the ICCF. Lag values are generally consistent among the three methods, though we typically measure smaller uncertainties with JAVELIN and CREAM than with the ICCF, given the more physically motivated light curve interpolation and more robust statistical modeling of the former two methods. The median redshift of our H beta-detected sample of quasars is 0.53, significantly higher than that of the previous reverberation mapping sample. We find that in most objects, the time delay of the Ha emission is consistent with or slightly longer than that of H beta. We measure black hole masses using our measured time delays and line widths for these quasars. These black hole mass measurements are mostly consistent with expectations based on the local M-BH-sigma* relationship, and are also consistent with single-epoch black hole mass measurements. This work increases the current sample size of reverberation-mapped active galaxies by about two-thirds and represents the first large sample of reverberation mapping observations beyond the local universe (z < 0.3).

If speech is acquired by a close-talking microphone in a controlled and noise-free environment, current state-of-the-art recognition systems often show an acceptable error rate. The use of close-talking microphones, however, may be too restrictive in many applications. Alternatively, distant-talking microphones, often placed several meters far from the speaker, may be used. Such setup is less intrusive, since the speaker does not have to wear any microphone, but the Automatic Speech Recognition (ASR) performance is strongly affected by noise and reverberation. The thesis is focused on ASR applications in a room environment, where reverberation is the dominant source of distortion, and considers both single- and multi-microphone setups. If speech is recorded in parallel by several microphones arbitrarily located in the room, the degree of distortion may vary from one channel to another. The difference among the signal quality of each recording may be even more evident if those microphones have different characteristics: some are hanging on the walls, others standing on the table, or others build in the personal communication devices of the people present in the room. In a scenario like that, the ASR system may benefit strongly if the signal with the highest quality is used for recognition. To find such signal, what is commonly referred as Channel Selection (CS), several techniques have been proposed, which are discussed in detail in this thesis. In fact, CS aims to rank the signals according to their quality from the ASR perspective. To create such ranking, a measure that either estimates the intrinsic quality of a given signal, or how well it fits the acoustic models of the recognition system is needed. In this thesis we provide an overview of the CS measures presented in the literature so far, and compare them experimentally. Several new techniques are introduced, that surpass the former techniques in terms of recognition accuracy and/or computational efficiency. A combination of different CS measures is also proposed to further increase the recognition accuracy, or to reduce the computational load without any significant performance loss. Besides, we show that CS may be used together with other robust ASR techniques, and that the recognition improvements are cumulative up to some extent. An online real-time version of the channel selection method based on the variance of the speech sub-band envelopes, which was developed in this thesis, was designed and implemented in a smart room environment. When evaluated in experiments with real distant-talking microphone recordings and with moving speakers, a significant recognition performance improvement was observed. Another contribution of this thesis, that does not require multiple microphones, was developed in cooperation with the colleagues from the chair of Multimedia Communications and Signal Processing at the University of Erlangen-Nuremberg, Erlangen, Germany. It deals with the problem of feature extraction within REMOS (REverberation MOdeling for Speech recognition), which is a generic framework for robust distant-talking speech recognition. In this framework, the use of conventional methods to obtain decorrelated feature vector coefficients, like the discrete cosine transform, is constrained by the inner optimization problem of REMOS, which may become unsolvable in a reasonable time. A new feature extraction method based on frequency filtering was proposed to avoid this problem.
Los actuales sistemas de reconocimiento del habla muestran a menudo una tasa de error aceptable si la voz es registrada por micr ofonos próximos a la boca del hablante, en un entorno controlado y libre de ruido. Sin embargo, el uso de estos micr ofonos puede ser demasiado restrictivo en muchas aplicaciones. Alternativamente, se pueden emplear micr ofonos distantes, los cuales a menudo se ubican a varios metros del hablante. Esta con guraci on es menos intrusiva ya que el hablante no tiene que llevar encima ning un micr ofono, pero el rendimiento del reconocimiento autom atico del habla (ASR, del ingl es Automatic Speech Recognition) en dicho caso se ve fuertemente afectado por el ruido y la reverberaci on. Esta tesis se enfoca a aplicaciones ASR en el entorno de una sala, donde la reverberaci on es la causa predominante de distorsi on y se considera tanto el caso de un solo micr ofono como el de m ultiples micr ofonos. Si el habla es grabada en paralelo por varios micr ofonos distribuidos arbitrariamente en la sala, el grado de distorsi on puede variar de un canal a otro. Las diferencias de calidad entre las señales grabadas pueden ser m as acentuadas si dichos micr ofonos muestran diferentes características y colocaciones: unos en las paredes, otros sobre la mesa, u otros integrados en los dispositivos de comunicaci on de las personas presentes en la sala. En dicho escenario el sistema ASR se puede bene ciar enormemente de la utilizaci on de la señal con mayor calidad para el reconocimiento. Para hallar dicha señal se han propuesto diversas t ecnicas, denominadas CS (del ingl es Channel Selection), las cuales se discuten detalladament en esta tesis. De hecho, la selecci on de canal busca ranquear las señales conforme a su calidad desde la perspectiva ASR. Para crear tal ranquin se necesita una medida que tanto estime la calidad intr nseca de una selal, como lo bien que esta se ajusta a los modelos ac usticos del sistema de reconocimiento. En esta tesis proporcionamos un resumen de las medidas CS hasta ahora presentadas en la literatura, compar andolas experimentalmente. Diversas nuevas t ecnicas son presentadas que superan las t ecnicas iniciales en cuanto a exactitud de reconocimiento y/o e ciencia computacional. Tambi en se propone una combinaci on de diferentes medidas CS para incrementar la exactitud de reconocimiento, o para reducir la carga computacional sin ninguna p erdida signi cativa de rendimiento. Adem as mostramos que la CS puede ser empleada junto con otras t ecnicas robustas de ASR, tales como matched condition training o la normalizaci on de la varianza y la media, y que las mejoras de reconocimiento de ambas aproximaciones son hasta cierto punto acumulativas. Una versi on online en tiempo real del m etodo de selecci on de canal basado en la varianza del speech sub-band envelopes, que fue desarrolladas en esta tesis, fue diseñada e implementada en una sala inteligente. Reportamos una mejora signi cativa en el rendimiento del reconocimiento al evaluar experimentalmente grabaciones reales de micr ofonos no pr oximos a la boca con hablantes en movimiento. La otra contribuci on de esta tesis, que no requiere m ultiples micr ofonos, fue desarrollada en colaboraci on con los colegas del departamento de Comunicaciones Multimedia y Procesamiento de Señales de la Universidad de Erlangen-Nuremberg, Erlangen, Alemania. Trata sobre el problema de extracci on de caracter sticas en REMOS (del ingl es REverberation MOdeling for Speech recognition). REMOS es un marco conceptual gen erico para el reconocimiento robusto del habla con micr ofonos lejanos. El uso de los m etodos convencionales para obtener los elementos decorrelados del vector de caracter sticas, como la transformada coseno discreta, est a limitado por el problema de optimizaci on inherente a REMOS, lo que har a que, utilizando las herramientas convencionales, se volviese un problema irresoluble en un tiempo razonable. Para resolver este problema hemos desarrollado un nuevo m etodo de extracci on de caracter sticas basado en fi ltrado frecuencial
Els sistemes actuals de reconeixement de la parla mostren sovint una taxa d'error acceptable si la veu es registrada amb micr ofons pr oxims a la boca del parlant, en un entorn controlat i lliure de soroll. No obstant, l' us d'aquests micr ofons pot ser massa restrictiu en moltes aplicacions. Alternativament, es poden utilitzar micr ofons distants, els quals sovint s on ubicats a diversos metres del parlant. Aquesta con guraci o es menys intrusiva, ja que el parlant no ha de portar a sobre cap micr ofon, per o el rendiment del reconeixement autom atic de la parla (ASR, de l'angl es Automatic Speech Recognition) en aquest cas es veu fortament afectat pel soroll i la reverberaci o. Aquesta tesi s'enfoca a aplicacions ASR en un ambient de sala, on la reverberaci o es la causa predominant de distorsi o i es considera tant el cas d'un sol micr ofon com el de m ultiples micr ofons. Si la parla es gravada en paral lel per diversos micr ofons distribuï ts arbitràriament a la sala, el grau de distorsi o pot variar d'un canal a l'altre. Les difer encies en qualitat entre els senyals enregistrats poden ser m es accentuades si els micr ofons tenen diferents caracter stiques i col locacions: uns a les parets, altres sobre la taula, o b e altres integrats en els aparells de comunicaci o de les persones presents a la sala. En un escenari com aquest, el sistema ASR es pot bene ciar enormement de l'utilitzaci o del senyal de m es qualitat per al reconeixement. Per a trobar aquest senyal s'han proposat diverses t ecniques, anomenades CS (de l'angl es Channel Selection), les quals es discuteixen detalladament en aquesta tesi. De fet, la selecci o de canal busca ordenar els senyals conforme a la seva qualitat des de la perspectiva ASR. Per crear tal r anquing es necessita una mesura que estimi la qualitat intr nseca d'un senyal, o b e una que valori com de b e aquest s'ajusta als models ac ustics del sistema de reconeixement. En aquesta tesi proporcionem un resum de les mesures CS ns ara presentades en la literatura, comparant-les experimentalment. A m es, es presenten diverses noves t ecniques que superen les anteriors en termes d'exactitud de reconeixement i / o e ci encia computacional. Tamb e es proposa una combinaci o de diferents mesures CS amb l'objectiu d'incrementar l'exactitud del reconeixement, o per reduir la c arrega computacional sense cap p erdua signi cativa de rendiment. A m es mostrem que la CS pot ser utilitzada juntament amb altres t ecniques robustes d'ASR, com ara matched condition training o la normalitzaci o de la varian ca i la mitjana, i que les millores de reconeixement de les dues aproximacions s on ns a cert punt acumulatives. Una versi o online en temps real del m etode de selecci o de canal basat en la varian ca de les envolvents sub-banda de la parla, desenvolupada en aquesta tesi, va ser dissenyada i implementada en una sala intel ligent. A l'hora d'avaluar experimentalment gravacions reals de micr ofons no pr oxims a la boca amb parlants en moviment, es va observar una millora signi cativa en el rendiment del reconeixement. L'altra contribuci o d'aquesta tesi, que no requereix m ultiples micr ofons, va ser desenvolupada en col laboraci o amb els col legues del departament de Comunicacions Multimedia i Processament de Senyals de la Universitat de Erlangen-Nuremberg, Erlangen, Alemanya. Tracta sobre el problema d'extracci o de caracter stiques a REMOS (de l'angl es REverberation MOdeling for Speech recognition). REMOS es un marc conceptual gen eric per al reconeixement robust de la parla amb micr ofons llunyans. L' us dels m etodes convencionals per obtenir els elements decorrelats del vector de caracter stiques, com ara la transformada cosinus discreta, est a limitat pel problema d'optimitzaci o inherent a REMOS. Aquest faria que, utilitzant les eines convencionals, es torn es un problema irresoluble en un temps raonable. Per resoldre aquest problema hem desenvolupat un nou m etode d'extracci o de caracter ístiques basat en fi ltrat frecuencial.

Thesis (M.S. in Applied Physics) Naval Postgraduate School, September 1999.
"September 1999". Thesis advisor(s):, Andrés Larraza, Bruce C. Denardo. Includes bibliographical references (p. 49). Also available online.

Synthetic speech is a valuable means of output, in a range of application contexts, for people with visual, cognitive, or other impairments or for situations were other means are not practicable. Noise and reverberation occur in many of these application contexts and are known to have devastating effects on the intelligibility of natural speech, yet very little was known about the effects on synthetic speech based on unit selection or hidden Markov models. In this thesis, we put forward an approach for assessing the intelligibility of synthetic and natural speech in noise, reverberation, or a combination of the two. The approach uses an experimental methodology consisting of Amazon Mechanical Turk, Matrix sentences, and noises that approximate the real-world, evaluated with generalized linear mixed models. The experimental methodologies were assessed against their traditional counterparts and were found to provide a number of additional benefits, whilst maintaining equivalent measures of relative performance. Subsequent experiments were carried out to establish the efficacy of the approach in measuring intelligibility in noise and then reverberation. Finally, the approach was applied to natural speech and the two synthetic speech systems in combinations of noise and reverberation. We have examine and report on the intelligibility of current synthesis systems in real-life noises and reverberation using techniques that bridge the gap between the audiology and speech synthesis communities and using Amazon Mechanical Turk. In the process, we establish Amazon Mechanical Turk and Matrix sentences as valuable tools in the assessment of synthetic speech intelligibility.

This thesis probes the performance of one-channel time-reversal acoustics in a chamber in terms of the geometry of the cavity. In particular, a rectangular chamber is compared to an enclosure that has a stadium shape. The mode structure in the rectangular cavity is highly symmetric, while it is highly irregular in the stadium-shaped cavity. Time- reversal acoustic techniques produce an improved focus in the latter. The focusing quality is determined as a function of frequency, time-reversal window size, and spatial extent. A scheme for encrypted acoustic communication, both in air and underwater, that uses multiple broadband signals with identical bandwidth, Hanning window source spectra, and center frequencies separated by half the bandwidth, allowing for null detection between adjacent signals, is successfully investigated.

Communication often takes place in reverberant spaces making it harder for listeners to understand speech. In such difficult environments, listeners would benefit from being able to locate the sound source. In noisy or reverberant environments hearing-aid wearers often complain that their aids do not sufficiently help to understand speech or to localise a sound source. Simple amplification does not fully resolve the problem and sometimes makes it worse. Recent improvements in hearing aids, such as compression and filtering, can significantly alter the Interaural Time Difference (ITD) and the Inter-aural Level Difference (ILD) cues. Digital signal processing also tends to restrict the availability of fine structure cues, thereby forcing the listener to rely on envelope and level cues. The effect of digital signal processing on localisation, as felt by hearing aid wearers in different listening environments, is not well investigated. In this thesis, we aimed to investigate the effect of reverberation on localisation performance of normal hearing and hearing impaired listeners, and to determine the effects that hearing aids have on localisation cues. Three sets of experiments were conducted: in the first set (n=22 normal hearing listeners) results showed that the participants’ sound localisation ability in simulated reverberant environments is not significantly different from performance in a real reverberation chamber. In the second set of four experiments (n=16 normal hearing listeners), sound localisation ability was tested by introducing simulated reverberation and varying signal onset/offset times of different stimuli – i.e. speech, high-pass speech, low-pass speech, pink noise, 4 kHz pure tone, and 500 Hz pure tone. In the third set of experiments (n=28 bilateral Siemens Prisma 2 Pro hearing aid users) we investigated aided and unaided localisation ability of hearing impaired listeners in anechoic and simulated reverberant environments. Participants were seated in the middle of 21 loudspeakers that were arranged in a frontal horizontal arc (180°) in an anechoic chamber. Simulated reverberation was presented from four corner-speakers. We also performed physical measurements of ITDs and ILDs using a KEMAR simulator. Normal hearing listeners were not significantly affected in their ability to localise speech and pink noise stimuli in reverberation, however reverberation did have a significant effect on localising a 500 Hz pure tone. Hearing impaired listeners performed consistently worse in all simulated reverberant conditions. However, performance for speech stimuli was only significantly worse in the aided conditions. Unaided hearing impaired listeners showed decreased performance in simulated reverberation, specifically, when sounds came from lateral directions. Moreover, low-pass pink noise was most affected by simulated reverberation both in aided and unaided conditions, indicating that reverberation mainly affects ITD cues. Hearing impaired listeners performed significantly worse in all conditions when using their hearing aids. Physical measurements and psychoacoustic experiments consistently indicated that amplification mainly affected the ILD cues. We concluded that reverberation destroys the fine structure ITD cues in sound signals to some extent, thereby reducing localisation performance of hearing impaired listeners for low frequency stimuli. Furthermore we found that hearing aid compression affects ILD cues, which impairs the ability of hearing impaired listener to localise a sound source. Aided sound localisation could be improved for bilateral hearing aid users, if the aids would synchronize compression between both sides.

L'intérêt de l'industrie automobile vers les chambres réverbérantes pour des essais en compatibilité électromagnétique a augmenté considérablement dans les dernières années. Cet intérêt vient de la possibilité d'exploiter certaines propriétés des chambres réverbérantes pour optimiser le procédé industriel de validation CEM. Cependant, plusieurs aspects théoriques qui sont à la base d'une méthodologie de test en chambre réverbérante restent à être explorés ou clarifiés. Certains de ces aspects sont approfondis dans cette thèse, dans l'optique des tests en immunité rayonnée pour des applications automobiles. La qualification de l'environnement électromagnétique d'une chambre réverbérante est d'abord considérée, avec une attention particulière vers l'incertitude de mesure et l'effet de charge du a l'introduction d'une voiture dans la chambre. Une approche statistique originale pour modéliser le couplage des champs avec des objets électriques est ensuite proposée et appliquée à l'analyse du couplage du champ avec le réseau électrique d'un véhicule. La reproductibilité des essais et la corrélation avec les résultats obtenus en chambre anéchoique sont enfin étudiées par voie théorique et par voie expérimentale au moyen de la répétition des essais effectués sur un dispositif électronique dans plusieurs chambres réverbérantes et dans une chambre semi-anéchoique.

Over the past ten years, Reverberation Chambers (RC) have emerged as a promising facility for the measurement of antenna characteristics for the wireless communications industry. The RC has begun to diverge from its initial purpose of performing Electromagnetic Compatibility (EMC) measurements, as conceived by H. A. Mendes back in 1968. Unlike the Anechoic Chamber (AC) however, the RC facility and measurement protocols are relatively in their infancy whose standardisation is yet to be finalised. The purpose of this thesis is to present a further study into reverberation chambers and their associated measurement procedures, aimed at smoothing the transition into a fully standardised and accepted facility within the measurement industry. This thesis is comprised of three main investigation areas. The first area under investigation concerns single port textile antennas designed for use in the on-body communications role. The purpose of this investigation is twofold: (1) to chart the efficiency and impedance matching performance of the antennas in both free space and on-body roles to completely characterise their performance, and (2) to devise and configure operational procedures for the measurement of antenna characteristics on human subjects using the RC. Two separate varieties of textile antenna are subject to investigation; the first consists of two antennas with an equally small ground plane designed for use in the Industrial Scientific and Medical (ISM) 2.45 GHz band. The second variety also consists of two separate antennas that have a larger ground plane size and are designed to offer a dual band characteristic; to operate at 2.45 GHz and 5.2 GHz respectively. The results for the smaller ground plane sized antennas show that in free space conditions, the textile antenna constructed from the higher conductivity textile material exhibits a greater level of efficiency which is expected. However, when placed on-body, the antenna with the lower conductivity textile material remarkably outperforms the antenna with the higher conductivity material which is contrary to expectations; this represents new and important knowledge. The results for the larger ground plane sized textile antennas conform completely to expectations. That is, the higher conductivity material outperformed the lower conductivity material in both free space and on-body roles. Comparing both cases, further new knowledge can be concluded in the fact that in addition to the conductivity of the textile material, the ground plane dimensions is also of crucial importance. The second area under investigation concerns multiport (array) antennas. This area is sub-divided into two sections to chart a distinction between multiport antennas designed for Multiple Input Multiple Output (MIMO) applications and more conventional array antennas that are not. The first section concerns the complete practical verification of two new dual feed Planar Inverted F Antennas (PIFAs). Results show that both antennas yield a high level of diversity gain and channel capacity (close to the theoretical maximums) and very low correlation between the two feeds despite the antennas small size. Furthermore, the antennas are also proved to be highly efficient at the desired frequency of operation. Comparing all performance results, it is possible to conclude that due to the small size and excellent performance of the new designs, they could be useful in more practical and commercial applications than larger sized elements that currently exist. The second section focuses upon more conventional larger sized array antennas used for radio astronomy applications. In this thesis, a series of power dividers is used to emulate a realistic ‘all - excited’ scenario, but the power divider approach has a consequence in that it will give rise to an external power loss that is not attributed to the antenna array. A new equation is developed in this work that allows for the accurate efficiency determination of the array and the de-embedding of the power divider in one. It is shown that the new equation can make this whole process simpler and straightforward to accomplish whilst maintaining accuracy. The final area under investigation concerns the design of reverberation chambers. The most common of the mode stirring techniques used in reverberation chambers is via the rotation or movement of electrically large metallic paddles inside the chamber known as ‘Mechanical Stirring’. In this thesis, a technique based upon a meanderline principle is used to cut slots into the mechanical stirring paddles to increase the current path length (induced when a wave hits the metallic surface) and thereby increase the electrical size of the paddle. New paddle designs for reverberation chambers are designed and verified. It is shown that the overall paddle dimensions do not need to be increased in size, meaning that the working volume of the chamber can remain as large as possible. The results show that the new designs exhibit enhanced performance over and above conventional paddle designs at lower modal numbers, meaning that any chamber will be able to better perform at frequencies where fewer modes exist. Results also show that at higher frequencies, the slot cuts do not adversely affect the chambers higher frequency performance. This work therefore has the potential to forge a new way of thinking when it comes to the design of mechanical stirrers in RC’s.

This master thesis examines the effects on the acoustical properties of a concert hall caused by an artificial reverberation system (ARS) and the possibility of simulating these properties. By examining the case of the Stockholm Concert Hall, which recently installed such a system, a greater understanding of the ARS will be gained and additional improvements of simulating such systems will be explored. This study comprises two parts: (1) objective data obtained through acoustical measurements are evaluated both internally and to other halls and (2) by computer simulation of the concert hall and its electronic reverberation system evaluate the acoustics of the hall. The study shows that the effect of the ARS on the acoustical properties of Stockholm Concert Hall is not excessive but noticeable. An 0.3 second increase in reverberation time is a desirable outcome but comes at the cost of clarity, which sees a reduction of 0.7 decibels. Moreover, it is possible to simulate a concert hall, having an ARS installed, with fairly realistic results. However, in order to compile the simulated impulse response, a script had to be created -combining the transfer functions related to each component of the reverberation chain from source to receiver, including all the microphones and loudspeakers of the ARS.
Det här examensarbetet undersöker påverkan på de akustiska egenskaperna hos en konsertlokal orsakad av ett artificiellt efterklangssystem. Likaså undersöks möjligheterna för att simulera dessa akustiska egenskaper. Genom att undersöka Stockholms konserthus, som nyligen installerade ett efterklangssystem, kommer en bättre förståelse för artificiella efterklangssystem skapas och ytterligare förbättringar för simulering kommer att möjliggöras. Den här studien genomförs i två delar: (1) objektiv data, inhämtad från akustiska mätningar, utvärderas både internt och mot andra konsertlokaler samt (2) genom datorsimulering av konsertlokalen och det elektroniska efterklangssystemet utvärderas de akustiska egenskaperna. Studien visar att inverkan på de akustiska egenskaperna hos Stockholms konserthus orsakade av det artificiella efterklangssystemet inte är överdrivna men noterbara. En önskad ökning av efterklangstiden med 0.3 sekunder uppnås men detta på bekostnad av att ljudets klarhet minskar med 0.7 decibel. Vidare är det möjligt att simulera ljudutbredningen i en konsertlokal som har ett efterklangssystem installerat med ett tämligen realistiskt resultat. För att uppnå detta simuleringsresultat skapas ett skript vilket väger samman alla överföringsfunktioner mellan ljudkällan och mottagaren, inklusive de mellan efterklangssystemets mikrofoner och högtalare.

Speech enhancement systems achieving a joint suppression of reverberation and background noise can be used in digital hearing aids, voice controlled systems or hands-free telephony. Demanding requirements for computational complexity, signal delay and speech quality must be fulfilled in order to achieve a satisfactory performance. The speech quality depends on how accurate the reverberation characteristics such as the reverberation time or the spectral variance of the late reverberant speech are estimated. In this thesis, an efficient algorithm for a blind reverberation time estimation based on maximum likelihood approach is introduced. The new algorithm allows to estimate reverberation times from a much wider range with acceptable accuracy. Variance of the late reverberant speech is another important quantity in dereverberation systems. Two late reverberant spectral variance estimation methods are compared with regard to estimation accuracy and computational complexity. Finally, the performance of the considered speech enhancement system is analyzed with the improved reverberation time estimator.

Since the last few years, the unprecedented growth of communication systems involving the propagation of electromagnetic waves is particularly due to developments in mobile phone technology. The reverberation chamber is a reliable bench-test, enabling the study of the effects of electromagnetic waves on a specific electronic appliance. However, the operating of a reverberation chamber being rather complicated, development of numerical models are of utmost importance to determine the crucial parameters to be considered.This thesis consists in the modelling and the simulation of the operating principles of a reverberation chamber by means of the Finite-Difference Time-Domain method. After a brief study based on field and power measurements performed in a reverberation chamber, the second chapter deals with the different problems encountered during the modelling. The consideration of losses being a very important factor in the operating of the chamber, two methods of implementation of these losses are set out in this chapter. Chapter~3 consists in the analysis of the influence of the stirrer on the first eigenmodes of the chamber; the latter modes can undergo a frequency shift of several MHz. Chapter~4 shows a comparison of results issued from high frequency simulations and theoretical statistical results. The problem of an object placed in the chamber, resulting in a field disturbance is also tackled. Finally, in the fifth chapter, a comparison of statistical results for stirrers having different shapes is set out.

This thesis explores compensation for reverberation in human listeners and machines. Late reverberation is typically understood as a distortion which degrades intelligibility. Recent research, however, shows that late reverberation is not always detrimental to human speech perception. At times, prolonged exposure to reverberation can provide a helpful acoustic context which improves identification of reverberant speech sounds. The physiology underpinning our robustness to reverberation has not yet been elucidated, but is speculated in this thesis to include efferent processes which have previously been shown to improve discrimination of noisy speech. These efferent pathways descend from higher auditory centres, effectively recalibrating the encoding of sound in the cochlea. Moreover, this thesis proposes that efferent-inspired computational models based on psychoacoustic principles may also improve performance for machine listening systems in reverberant environments. A candidate model for perceptual compensation for reverberation is proposed in which efferent suppression derives from the level of reverberation detected in the simulated auditory nerve response. The model simulates human performance in a phoneme-continuum identification task under a range of reverberant conditions, where a synthetically controlled test-word and its surrounding context phrase are independently reverberated. Addressing questions which arose from the model, a series of perceptual experiments used naturally spoken speech materials to investigate aspects of the psychoacoustic mechanism underpinning compensation. These experiments demonstrate a monaural compensation mechanism that is influenced by both the preceding context (which need not be intelligible speech) and by the test-word itself, and which depends on the time-direction of reverberation. Compensation was shown to act rapidly (within a second or so), indicating a monaural mechanism that is likely to be effective in everyday listening. Finally, the implications of these findings for the future development of computational models of auditory perception are considered.

The extremely high absorption of energy of electromagnetic waves in underwater environments restricts the range of signals to be used to acoustic signals. In addition the sea is a complex medium in which many kinds of environmental changes, mul­tipath propagation phenomenon, masking of the signals of interest by noise and/or reverberation signals, and attenuation, among others, will affect the propagation of sound through it. On one hand, environmental changes will cause different degrees of nonstationarity at the signals to be processed. On the other hand, the use of acoustic waves will imply that, for the active sonar case, different Doppler shifts of the signals to track will take place as the relative radial velocity of the sonar platform to the contact varies. This will cause that in some instances the contact signals share not only time, but also frequency bins with the noise and/or the reverberation signals. For the noise-limited case, an optimum solution for signal detection based on the correlation receiver or Matched-filter, exists. However, for reverberation-limited environments there is not any optimum solution which is feasible to be implemented in a practical system. Adaptive filters grew out of the demand of systems capable of operating in uncertain, time-varying environments. Due to the wide range of applications for which they have shown to be useful, considerable amount of work has been dedicated during the last few years to their development. The preliminary part of the thesis presents a basic model of the underwater environment for the active sonar case upon which the suitability of certain adaptive structures for active echo detection and rang­ing is initially based. A classification and the description of some existing adaptive systems and their main characteristics are presented too. Subsequent parts of the thesis include the theoretical development of a generic adaptive algorithm which will operate with complex data sequences. Several sets of experiments are carried out and the results presented in order to investigate the suitability for the application of interest of several adaptive systems and algorithms. Adaptive processing the received signals as presented here must be understood as a preprocessing stage of the overall active sound navigation and ranging (sonar) problem. The study is restricted to the narrowband case.

Reverberation is a natural phenomenon observed in enclosed environments. It occurs due to the reflection of the signal from the walls and objects in the room. For humans, reverberation is beneficial as it reinforces sound and also provide the sensation of space. However, for automatic speech recognition even moderate amount of reverberation is very harmful. It corrupts the clean speech which leads to deterioration in the performance of the speech recognizer. Moreover, in the enclosed environment, reverberation has the most damaging affect over the accuracy of the recognizer. In literature, to improve speech recognition performance against environmental artifacts mostly noise compensation techniques have been proposed. As a consequence, the problem of reverberation has received relatively less attention. Lately, some techniques have emerged which are specifically tailored for compensating the effects of reverberation. Nevertheless, the problem of reverberation is far from being solved. Therefore, to handle reverberation and provide robustness to speech recognition, we propose "Semi-blind adaptation" technique which adapts the clean acoustic models to the reverberant environment and thus provide improved performance. Semi-blind adaptation technique works in two phases, in the first phase reverberation model is estimated and in the second phase using the reverberation model, adaptation of the clean acoustic models is performed. The reverberation model (Pw-EDC) proposed in this technique models the non-diffuse nature of the rooms. Therefore, the Pw-EDC model has dual slope energy decay where the first slope represents the steep decay of early reflections and second slope represents the slow decay of late reflections. The parameters to model early reflections decay were empirically calculated and to find the parameter of late reflections decay we proposed Gaussian mixture models (GMMs) based reverberation time estimation technique. Late reflections decay parameter is estimated by first training a pool of GMMs where each model represents the reverberation time of the data on which it is trained. In the test phase, test data is matched with these models and the GMM which matches with highest probability provide the estimate of late reflections decay parameter. To adapt the acoustic models, reverberation energy contributions are estimated by using the Pw-EDC model. The parameters of the current state in the model (i.e., only means) are adapted by adding the reverberation energy contributions of the previous states to the current state. In this manner, the dispersion of energy caused by the reverberation is compensated. Adaptation is performed not only on static parameters but also on dynamic parameters of the model. After adaptation, the models are evaluated on data from low, medium and high reverberant environments. The efficacy of the proposed adaptation technique is evaluated on small and medium vocabulary tasks. For these tasks reverberant data is generated by convolving clean signals with impulse responses taken from SIREAC and AIR databases. SIREAC provides RIRs of office and living room and it also has a facility to modify the reverberation time. Therefore, in our experiments the reverberation time of the RIRs is varied from 200 to 900 ms in steps of 100 ms for both rooms of SIREAC. In AIR environment, RIRs are obtained from studio booth, meeting, office and lecture rooms. These rooms have very low, low, medium and high reverberation times respectively. For small vocabulary task, Pw-EDC adaptation provide considerable improvements compared to the baseline results especially at medium and high reverberation times in both environments. Pw-EDC adaptation is compared with contemporary adaptation technique (Exp-EDC adaptation) which also adapts the models in the same manner, except it uses a crude reverberation model. It was found, Pw-EDC adaptation gives better performance in all the rooms of both environments. Pw-EDC is also compared with state-of-the-art adaptation technique i.e., unsupervised MLLR and it was found that Pw-EDC adaptation provide similar performance to MLLR only when the models contain static coefficients. For medium vocabulary task, Pw-EDC adaptation provide better performance than Exp-EDC adaptation in both the environments. However, when compared against unsupervised MLLR it shows relatively poor performance. The reason for such dismal performance is the inaccurate adaptation of dynamic coefficients of the models. In the end, the robustness of proposed adaptation technique is due to the precise modeling and estimation of reverberation energy decay by Pw-EDC model. Using Pw-EDC model, the semi-blind adaptation has shown consistent improvements across low, medium and high reverberant environments in both small and medium vocabulary speech recognition task.

Thesis (M. S.)--Mechanical Engineering, Georgia Institute of Technology, 2005.
Berthelot, Yves, Committee Member ; Cunefare, Kenneth A, Committee Chair ; Lynch, Christopher, Committee Member. Includes bibliographical references.

This master thesis project was carried out during the spring semester of 2016 at the company Limes Audio. The company specializes in making software, electronics, industrial design and mechanics with the aim to improve audio quality in loudspeaking communication systems. The performance of audio conferencing systems may degrade if there are distortions, or if the acoustical properties of the room are unfavorable. To ensure that the system works optimally, any unwanted effects must first be identified. This thesis will cover the necessary theory of acoustical systems and measurements. The implementations of three different measurement sequences are presented. The measurements are evaluated on three different speakerphone units and three venues to assess the accuracy of the presented methods. The results indicate that it is possible to use a single measurement signal, the exponential sine sweep, to measure both room acoustical reverberation time and loudspeaker distortion for a speakerphone setup.

Thesis (MTech (Electrical Engineering))--Cape Peninsula University of Technology, 2017.
This research project focuses on the implementation of a Reverberation Chamber (RC) by the transformation of an existing electromagnetically shielded room. The reverberation chamber is a kind of shielded room designed to create a statistically random internal electromagnetic environment. The reverberating environment makes it possible to obtain high field strengths from a relatively low input power. The electric fields in the chamber have to be stirred to achieve a statistically uniform field. The first part of this thesis presents an overview of reverberation chamber principles and preliminary calculations are done: the lowest usable frequency is estimated to be close to 300 MHz from empirical criteria. Modelling of the statistical environment is then presented, where electromagnetic quantities are characterised by probability density functions (Gaussian, Rayleigh and exponential); correlation issues are also presented. Measurements are performed in the frequency range of 800 MHz – 4 GHz, dictated by the antennas available for this research study. An investigation of cable losses is conducted, followed by a discussion on measurement accuracy. Mechanical stirrers are designed and manufactured. Electromechanical components are selected based on the literature study. Measurements are obtained through an automated setup using MATLAB®. To verify that the RC, with its in-house designed mechanical stirrers, is well-operated, the stirring ratio is experimentally determined. After this first test, an exhaustive investigation of probability density functions is conducted, taking into account correlation issues. Measurements show that the quality factor of the chamber is close to 2000 at 3 GHz, and that 60 independent stirrer positions at 4 GHz can be used for statistical analyses. Finally, the uniformity test is performed with an improved accuracy using frequency stirring. In conclusion, the CPUT RC passes the validation procedure according to the IEC 61000-4-21 standard by generating the required field uniformity within the accepted uncertainty level.

When capturing speech signals using a distant microphone within a confined acoustic space, the recordings are often degraded by reverberation. This can have a detrimental impact on the quality and intelligibility of speech, especially when combined with acoustic noise. In recent years, there has been increasing demand for effective ways of combating the damaging effects of reverberation in applications such as hands-free telephony or hearing-aids technology. However, the task of providing a blind single-channel dereverberation method robust to high levels of noise and suitable for real-time processing remains a challenge. An important prerequisite for many single-channel dereverberation algorithms is the estimation of the acoustic parameters governing reverberation. In this thesis, a novel online method of estimating these parameters jointly with the interfering signal powers is proposed that is based on a combination of Voice Activity Detection and Extended Kalman Filters. This method is then extended to take into account the spectral structure of clean speech signals and to perform dereverberation by applying a time-frequency gain to the degraded speech spectrogram. The estimation of this gain is formulated as a Bayesian filtering problem conditioned on a Hidden Markov Model. In order to evaluate the proposed algorithm in terms of speech intelligibility, a novel algorithm for measuring Psychometric Functions efficiently in listening experiments is presented. The algorithms developed are evaluated on both simulated and real recordings and are compared with existing state-of-the art alternatives.

Although it is widely accepted that the diffusion of early reflections in acoustic spaces intended for music performance greatly improves the perceived quality of sound, current manufacturers of synthetic reverberation engines continue to model reflecting surfaces as having almost perfectly specular characteristics. This dissertation describes a hybrid method of simulating diffusion based on both physical and phenomenological modeling components.
In 1979, Manfred Schroeder described a method of designing and constructing diffusing surfaces based on a rather simple mathematical algorithm which provides diffused reflections in predictable frequency bands. This structural device, now known as a "Schroeder diffuser," has become a standard geometry used in constructing diffusive surfaces for spaces intended for music rehearsal, recording and performance. While it is possible to use DSP to model the characteristics of reflections off such a surface, a reflection model based exclusively on a surface constructed of a Schroeder diffuser has proven in informal tests to be as aesthetically inadequate as a perfectly specular model. Control of both the spatial and temporal envelopes of the diffusive reflection are required by an end user in order to tailor the reflection characteristics to the desired impression.
In 1974 an empirical model for computing light reflections off objects in a three-dimensional environment was developed by Phong Bui-Toung. This algorithm incorporated both a specular and diffuse component with relationships controlled by an end user.
This dissertation describes the adaptation and implementation of the Phong shading algorithm in conjunction with a physical model of components of the Schroeder diffuser for the modeling of diffuse reflections in synthetic acoustic environments. The inclusion of the Phong algorithm provides precise control over the balance between the spectral and diffusive components of the reflection. In addition, directivity functions for sound sources and receivers in the virtual space are described.
Analysis and evaluation of the model using mathematical and empirical methodologies are discussed and stereo and multichannel audio examples produced by the system are included.

A reverberation chamber is a highly conductive cavity in which it is possible to generatehigh electromagnetic elds that can be considered statistically homogeneous. Reverberationchambers have existed as a resource for electromagnetic compatibility (EMC) testing formore than 30 years. Working to promote international co-operation on standardization, severalorganizations have published various EMC standards. At Combitech AB in Linkopinghave a chamber that is commercially used for dierent types of measurements. To makethe chamber more attractive and versatile it is within their interest to get a system whichis compatible with the latest standards. The project aimed to develop a control system forthe reverberation chamber at Combitech and to equip it with functionality enabling it tomake measurements according to current EMC standards. Using the programming softwareAgilent VEE a program was developed to communicate with the supporting equipment andmanage test routines. Within the program software lies functionality directly associate withmode stirring and mode tuning procedures for standards DO-160F and MIL-STD. Duringmeasurements the program has abilities for skipping frequencies, pause/continue the currentsweep, executing preset events and adding commented markers to the plot window. Someother usable functionality implemented is project save/load, help section, directory selectionand data export abilities. The system holds functionality enabling measurements accordingto the standards in question, though future work will be needed to be able to carry througha proper and correct measurement routine.

The objective of this thesis is the development of signal processing and analysis techniques that would provide sharply improved speech recognition accuracy in highly reverberant environments. Speech is a natural medium of communication for humans, and in the last decade various speech technologies like automatic speech recognition (ASR), voice response systems etc. have considerably matured. The above systems rely on the clarity of the captured speech but many of the real-world environments include noise and reverberation that mitigate the system performance. The key focus of the thesis is on the robustness of ASR to reverberation. In our work, we first provide a new framework to adequately and efficiently represent the problem of reverberation in speech feature domains. Although our framework incurs modeling approximation errors, we believe that it provides a good basis for developing reverberation compensation algorithms. Based on our framework, we successfully develop a number of dereverberation algorithms. The algorithms reduce the uncertainly involved in dereverberation tasks by using speech knowledge in terms of cepstral auto-correlation, cepstral distribution, and, non-negativity and sparsity of spectral values. We demonstrate the success of our algorithms on clean-training as well as matched-training. Apart from dereverberation, we also provide an approach for noise robustness via a temporal-difference operation in the speech spectral domain. There, via a theoretical analysis, we predict an expected improvement in the SNR threshold shift for whitenoise conditions. We also empirically quantify and study speech-feature level distortion with respect to speech-signal level additive noise. Finally, we provide a new framework for a joint reverberation and noise representation and compensation. The new framework generalizes the spectral domain reverberation framework by incorporating an additive noise term. Working under the new framework, we combine our dereverberation and noise compensation approaches for better dereverberation as well as for the most challenging speech recognition task that includes both noise and reverberation components.

Thesis: S.M., Joint Program in Applied Ocean Science and Engineering (Massachusetts Institute of Technology, Department of Mechanical Engineering; and the Woods Hole Oceanographic Institution), 2015.
Cataloged from PDF version of thesis.
Includes bibliographical references (page 77).
Due to the costs involved and time required to perform experiments at sea, it is important to provide accurate simulations of the ocean environment. Using the ray tracing code, BELLHOP, the Mission Oriented Operating Suite (MOOS), methods outlined by the Naval Research Laboratory (NRL) for bottom reverberation, and MATLAB, a model will be developed to incorporate the effects of bottom reverberation into the BELLHOP suite of code. This will be accomplished by using BELLHOP to generate a ray trace and eigen ray file. Then a MATLAB script will take the BELLHOP information and calculate the reverberation level using the NRL model by measuring the amplitude and reverberation at a receiver array simulated on the ocean floor. These reverberation values will then be used to determine the reverberation level at the source due to these bottom interactions. Testing of the simulation will include deep and shallow ocean profiles and multiple sound speed profiles (SSP). Following this testing, the goal is to implement the model in existing C++ code used for the testing of AUV systems. The ability to accurately model the ocean will not only allow for testing of autonomy code in the laboratory, but also make it possible to refine and calibrate code making ship time more efficient.
by Thomas Edward Miller.
S.M.