In teleconferencing the audio quality is key to conducting successful meetings. The conference room setting imposes various challenges on the speech signal processing, such as noise and interfering signals, reverberation, or participants positioned far from the telephone unit. This work aims at improving the received speech signal of a conference telephone by implementing sound source localization and beamforming. The implemented microphone array signal processing techniques are compared to the performance of an existing multi-microphone solution and evaluated under various conditions using a planar uniform circular array. Recordings of test-sequences for the evaluation were performed using a custom-built array mockup. The implemented algorithms did not show good enough performance to motivate the increased computational complexity compared to the existing solution. Moreover, an increase in number of microphones used was concluded to have little or no effect on the performance of the methods. The type of microphone used was, however, concluded to have impact on the performance and a subjective listening evaluation indicated a preference for omnidirectional microphones which is recommended to investigate further. / God ljudkvalitet är en grundsten för lyckade telefonmöten. Miljön i ett konferens-rum medför ett flertal olika utmaningar för behandlingen av mikrofonsignalerna: det kan t.ex. vara brus och störningar, eller att den som talar befinner sig långt från telefonen. Målet med detta arbete är att förbättra den talsignal som tas upp av en konferenstelefon genom att implementera lösningar för lokalisering av talaren och riktad ljudupptagning med hjälp av ett flertal mikrofoner. De implementerade metoderna jämförs med en befintlig lösning och utvärderas under olika brusscenarion för en likformig cirkulär mikrofonkonstellation. För utvärderingen användes testsignaler som spelades in med en specialbyggd enhet. De implementerade algoritmerna kunde inte uppvisa en tillräcklig förbättring i jämförelse med den befintliga lösningen för att motivera den ökade beräkningskomplexitet de skulle medföra. Dessutom konstaterades att en fördubbling av antalet mikrofoner gav liten eller ingen förbättring på metoderna. Vilken typ av mikrofon som användes konstaterades däremot påverka resultatet och en subjektiv utvärdering indikerade en preferens för de rundupptagande mikrofonerna, en skillnad som föreslås undersökas vidare.

Nas últimas décadas, a poluição sonora tornou-se um grande problema para a sociedade. É por esta razão que a indústria tem aumentado seus esforços para reduzir a emissão de ruído. Para fazer isso, é importante localizar quais partes das fontes sonoras são as que emitem maior energia acústica. Conhecer os pontos de emissão é necessário para ter o controle das mesmas e assim poder reduzir o impacto acústico-ambiental. Técnicas como \"beamforming\" e \"Near-Field Acoustic Holography\" (NAH) permitem a obtenção de imagens acústicas. Essas imagens são obtidas usando um arranjo de microfones localizado a uma distância relativa de uma fonte emissora de ruído. Uma vez adquiridos os dados experimentais pode-se obter a localização e magnitude dos principais pontos de emissão de ruído. Do mesmo modo, ajudam a localizar fontes aeroacústicas e vibro acústicas porque são ferramentas de propósito geral. Usualmente, estes tipos de fontes trabalham em diferentes faixas de frequência de emissão. Recentemente, foi desenvolvida a transformada de Kronecker para arranjos de microfones, a qual fornece uma redução significativa do custo computacional quando aplicada a diversos métodos de reconstrução de imagens, desde que os microfones estejam distribuídos em um arranjo separável. Este trabalho de mestrado propõe realizar medições com sinais reais, usando diversos algoritmos desenvolvidos anteriormente em uma tese de doutorado, quanto à qualidade do resultado obtido e à complexidade computacional, e o desenvolvimento de alternativas para tratamento de dados quando alguns microfones do arranjo apresentarem defeito. Para reduzir o impacto de falhas em microfones e manter a condição de que o arranjo seja separável, foi desenvolvida uma alternativa para utilizar os algoritmos rápidos, eliminando-se apenas os microfones com defeito, de maneira que os resultados finais serão obtidos levando-se em conta todos os microfones do arranjo. / In recent decades, noise pollution has become a major problem for society. It is for this reason that the industry has increased its efforts to reduce the emission of noise. To do this, it is important to find out which parts of the sound sources are emitting greater acoustic energy. Knowing the emission points is required to keep track of them and thus be able to reduce acoustic and environmental impact. Techniques such as \"beamforming\" and \"Near-Field Acoustic Holography\" (NAH) allow obtaining acoustic images. These images are obtained using an array of microphones located at a relative distance from a noise source. Once acquired experimental data, one can obtain the location and the magnitude of the main points of noise emission. Similarly, we can find aeroacoustic and vibroacoustic sources, because they are general-purpose tools. Usually, these types of sources work in different frequency bands of the emission. Recently, a Kronecker Array Transform (KAT) was developed to microphone arrays, which provides a significant reduction in computational cost when applied to various image reconstruction methods, provided that the microphones are distributed in a separable array. This thesis proposes perform measurements with real signals using different algorithms previously developed in a dissertation about the quality of the obtained results and computational complexity, and the development of alternatives for data processing when some microphones of the array are defective. To reduce the impact of failures in microphones and maintain the condition that the array is separable, one alternative has been developed to use the fast algorithms by removing only the defective microphones, so that the final results will be obtained taking into account every microphone of the array.

Nas últimas décadas, a poluição sonora tornou-se um grande problema para a sociedade. É por esta razão que a indústria tem aumentado seus esforços para reduzir a emissão de ruído. Para fazer isso, é importante localizar quais partes das fontes sonoras são as que emitem maior energia acústica. Conhecer os pontos de emissão é necessário para ter o controle das mesmas e assim poder reduzir o impacto acústico-ambiental. Técnicas como \"beamforming\" e \"Near-Field Acoustic Holography\" (NAH) permitem a obtenção de imagens acústicas. Essas imagens são obtidas usando um arranjo de microfones localizado a uma distância relativa de uma fonte emissora de ruído. Uma vez adquiridos os dados experimentais pode-se obter a localização e magnitude dos principais pontos de emissão de ruído. Do mesmo modo, ajudam a localizar fontes aeroacústicas e vibro acústicas porque são ferramentas de propósito geral. Usualmente, estes tipos de fontes trabalham em diferentes faixas de frequência de emissão. Recentemente, foi desenvolvida a transformada de Kronecker para arranjos de microfones, a qual fornece uma redução significativa do custo computacional quando aplicada a diversos métodos de reconstrução de imagens, desde que os microfones estejam distribuídos em um arranjo separável. Este trabalho de mestrado propõe realizar medições com sinais reais, usando diversos algoritmos desenvolvidos anteriormente em uma tese de doutorado, quanto à qualidade do resultado obtido e à complexidade computacional, e o desenvolvimento de alternativas para tratamento de dados quando alguns microfones do arranjo apresentarem defeito. Para reduzir o impacto de falhas em microfones e manter a condição de que o arranjo seja separável, foi desenvolvida uma alternativa para utilizar os algoritmos rápidos, eliminando-se apenas os microfones com defeito, de maneira que os resultados finais serão obtidos levando-se em conta todos os microfones do arranjo. / In recent decades, noise pollution has become a major problem for society. It is for this reason that the industry has increased its efforts to reduce the emission of noise. To do this, it is important to find out which parts of the sound sources are emitting greater acoustic energy. Knowing the emission points is required to keep track of them and thus be able to reduce acoustic and environmental impact. Techniques such as \"beamforming\" and \"Near-Field Acoustic Holography\" (NAH) allow obtaining acoustic images. These images are obtained using an array of microphones located at a relative distance from a noise source. Once acquired experimental data, one can obtain the location and the magnitude of the main points of noise emission. Similarly, we can find aeroacoustic and vibroacoustic sources, because they are general-purpose tools. Usually, these types of sources work in different frequency bands of the emission. Recently, a Kronecker Array Transform (KAT) was developed to microphone arrays, which provides a significant reduction in computational cost when applied to various image reconstruction methods, provided that the microphones are distributed in a separable array. This thesis proposes perform measurements with real signals using different algorithms previously developed in a dissertation about the quality of the obtained results and computational complexity, and the development of alternatives for data processing when some microphones of the array are defective. To reduce the impact of failures in microphones and maintain the condition that the array is separable, one alternative has been developed to use the fast algorithms by removing only the defective microphones, so that the final results will be obtained taking into account every microphone of the array.

<p>In modern cars a lot of effort is put on reducing the background noise level. Despite these efforts it is often difficult for persons in the rear seat(s) to hear the persons in the front seat. This is partly due to the background noise, but also geometry and acoustics properties of the passenger compartment. </p><p>The aim of this thesis was to implement a speech enhancement system to increase the audibility between the driver and the rear passenger(s). The speech enhancement system should not affect the directivity of the speech or increase the background noise level. </p><p>A speech enhancement system has been implemented on a DSP in a test car. A microphone was placed in front of the driver to collect his/her speech. The microphone signal was bandpass filtered to remove the main part of the background noise and to avoid aliasing. The signal was delayed before it was sent out in the rear loudspeaker. The delay made the speech from the driver reaching the rear passenger before the sound the rear loudspeakers. This delay was enough to get the right directivity of the sound, i.e. making speech sounding as if it came from the driver instead of the rear loudspeakers. </p><p>In the thesis other methods to reduce background noise and get directivity of the sound were evaluated, but not implemented in the test car. The evaluations of the system showed that the audibility was increased. At the same time the background noise level was not noticeable increased. The work has been performed at A2 Acoustics AB in Linköping, during spring 2003.</p>

In modern cars a lot of effort is put on reducing the background noise level. Despite these efforts it is often difficult for persons in the rear seat(s) to hear the persons in the front seat. This is partly due to the background noise, but also geometry and acoustics properties of the passenger compartment. The aim of this thesis was to implement a speech enhancement system to increase the audibility between the driver and the rear passenger(s). The speech enhancement system should not affect the directivity of the speech or increase the background noise level. A speech enhancement system has been implemented on a DSP in a test car. A microphone was placed in front of the driver to collect his/her speech. The microphone signal was bandpass filtered to remove the main part of the background noise and to avoid aliasing. The signal was delayed before it was sent out in the rear loudspeaker. The delay made the speech from the driver reaching the rear passenger before the sound the rear loudspeakers. This delay was enough to get the right directivity of the sound, i.e. making speech sounding as if it came from the driver instead of the rear loudspeakers. In the thesis other methods to reduce background noise and get directivity of the sound were evaluated, but not implemented in the test car. The evaluations of the system showed that the audibility was increased. At the same time the background noise level was not noticeable increased. The work has been performed at A2 Acoustics AB in Linköping, during spring 2003.

The speech of a person speaking in a noisy environment can be enhanced through electronic beamforming using spatially distributed microphones. As this approach demands precise information about the microphone locations, its application is limited in places where microphones must be placed quickly or changed on a regular basis. Highly precise calibration or measurement process can be tedious and time consuming. In order to understand tolerable limits on the calibration process, the impact of microphone position error on the intelligibility is examined. Analytical expressions are derived by modeling the microphone position errors as a zero mean uniform distribution. Experiments and simulations were performed to show relationships between precision of the microphone location measurement and loss in intelligibility. A variety of microphone array configurations and distracting sources (other interfering speech and white noise) are considered. For speech near the threshold of intelligibility, the results show that microphone position errors with standard deviations less than 1.5cm can limit losses in intelligibility to within 10% of the maximum (perfect microphone placement) for all the microphone distributions examined. Of different array distributions experimented, the linear array tends to be more vulnerable whereas the non-uniform 3D array showed a robust performance to positional errors.

L'étude expérimentale de la propagation et du rayonnement multimodal en guide large est abordée via des mesures par antennerie microphonique de la réflexion des modes pour différentes extrémités. Le banc expérimental est constitué d'un guide large fermé à une extrémité et débouchant sur différentes terminaisons à l'autre extrémité ; en paroi du guide sont branchées une source acoustique et deux antennes microphoniques. Chaque composant du banc est étudié pour améliorer les résultats de mesure. Une méthode de vérification des performances des haut-parleurs constituant la source et une méthode de pilotage de la source acoustique sont proposées pour favoriser la génération des différents modes de propagation de l'onde. Une méthode de calibration in-situ pour l'antenne est développée pour les différents modes. Un calcul des incertitudes pour l'estimation du coefficient de réflexion est proposé.Enfin les mesures sont effectuées pour différentes extrémités de guide : avec une bride, sans épaisseur, avec un écran infini. Le principe de la méthode de mesure de la réflexion des différents modes consiste à appliquer la méthode du doublet microphonique adaptée aux signaux issus de la décomposition modale obtenue au moyen de deux antennes de microphones. Les résultats de mesure pour le mode plan sont avantageusement comparés aux résultats théoriques issus de la littérature. Les résultats pour les premiers modes supérieurs montrent l'aptitude du système à extraire le coefficient de réflexion en module et en phase suffisamment précisément pour distinguer l'effet de la condition de rayonnement. / The experimental study of multimodal propagation and radiation in a wide guide is proposed via measurements of the reflection of modes for different terminations by using microphone arrays. The experimental bench consists of a wide guide closed at one end and ended with different terminations at the other end; an acoustic source and two microphone arrays are flush-mounted to the wall of the guide. Each component of the bench is first studied to improve the measurement results. A method of verifying the performance of the loudspeakers constituting the acoustic source and a method of controlling the acoustic source are proposed in order to facilitate the generation of the different modes of propagation of the wave. An in-situ calibration method for the microphone array is developed for the different modes. A calculation of the uncertainties for the estimation of the reflection coefficient is proposed.Then, measurements are performed for different guide terminations: with a finite flange, without flange, and with an infinite flange. The measurement of the reflection for the different modes consists of applying the method of two microphones to the signals from the modal decomposition obtained by means of two microphone arrays. Results of measurements for the plane mode are satisfactorily compared with theoretical results from the literature. Results for the first higher modes show the ability of the system to extract the reflection coefficient in modulus and in phase with sufficient precision to distinguish the effect of the radiation condition.

This thesis is related to the field of acoustic signal processing and its applications to emerging communication environments. Acoustic signal processing is a very wide research area covering the design of signal processing algorithms involving one or several acoustic signals to perform a given task, such as locating the sound source that originated the acquired signals, improving their signal to noise ratio, separating signals of interest from a set of interfering sources or recognizing the type of source and the content of the message. Among the above tasks, Sound Source localization (SSL) and Automatic Speech Recognition (ASR) have been specially addressed in this thesis. In fact, the localization of sound sources in a room has received a lot of attention in the last decades. Most real-word microphone array applications require the localization of one or more active sound sources in adverse environments (low signal-to-noise ratio and high reverberation). Some of these applications are teleconferencing systems, video-gaming, autonomous robots, remote surveillance, hands-free speech acquisition, etc. Indeed, performing robust sound source localization under high noise and reverberation is a very challenging task. One of the most well-known algorithms for source localization in noisy and reverberant environments is the Steered Response Power - Phase Transform (SRP-PHAT) algorithm, which constitutes the baseline framework for the contributions proposed in this thesis. Another challenge in the design of SSL algorithms is to achieve real-time performance and high localization accuracy with a reasonable number of microphones and limited computational resources. Although the SRP-PHAT algorithm has been shown to be an effective localization algorithm for real-world environments, its practical implementation is usually based on a costly fine grid-search procedure, making the computational cost of the method a real issue. In this context, several modifications and optimizations have been proposed to improve its performance and applicability. An effective strategy that extends the conventional SRP-PHAT functional is presented in this thesis. This approach performs a full exploration of the sampled space rather than computing the SRP at discrete spatial positions, increasing its robustness and allowing for a coarser spatial grid that reduces the computational cost required in a practical implementation with a small hardware cost (reduced number of microphones). This strategy allows to implement real-time applications based on location information, such as automatic camera steering or the detection of speech/non-speech fragments in advanced videoconferencing systems. As stated before, besides the contributions related to SSL, this thesis is also related to the field of ASR. This technology allows a computer or electronic device to identify the words spoken by a person so that the message can be stored or processed in a useful way. ASR is used on a day-to-day basis in a number of applications and services such as natural human-machine interfaces, dictation systems, electronic translators and automatic information desks. However, there are still some challenges to be solved. A major problem in ASR is to recognize people speaking in a room by using distant microphones. In distant-speech recognition, the microphone does not only receive the direct path signal, but also delayed replicas as a result of multi-path propagation. Moreover, there are multiple situations in teleconferencing meetings when multiple speakers talk simultaneously. In this context, when multiple speaker signals are present, Sound Source Separation (SSS) methods can be successfully employed to improve ASR performance in multi-source scenarios. This is the motivation behind the training method for multiple talk situations proposed in this thesis. This training, which is based on a robust transformed model constructed from separated speech in diverse acoustic environments, makes use of a SSS method as a speech enhancement stage that suppresses the unwanted interferences. The combination of source separation and this specific training has been explored and evaluated under different acoustical conditions, leading to improvements of up to a 35% in ASR performance. / Martí Guerola, A. (2013). Multichannel audio processing for speaker localization, separation and enhancement [Tesis doctoral no publicada]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/33101 / TESIS

This thesis explores problematics of speaker localization using microphone array. Aim of this thesis is implementation of algorithms for speaker localization and experiments with those algorithms. Calculation of TDOA was done using cross-correlation and hyperbolic method was used to calculate position estimation. Finished microphone array is able to locate speaker within certain variance. Results of this thesis allow reader to make assumptions regarding accuracy of localisation using microphone array and ARM kit with limited performance. Precision of position estimation using microphone array reached several decimeters, but this precision is dependent on distance from microphone array.

This master’s thesis deals with problems of noise source identification using nearfield acoustical holography (NAH). In the beginning there is the summary of basic terms and values of a sound pressure field, which is unnecessary for understanding of the theme. In the next part the thesis continues with more detailed description of the NAH technology and the historical context of its emergence. Measurement equipment which is used for scanning of sound pressure fields is also introduced. In addition, the kinds of NAH (according the shape of the wave front) are showed and the planar NAH is descripted most closely. Because of the NAH algorithms are implemented in the wave number domain (k-space), there is also a chapter focused to this problem in the thesis. There are briefly descripted some similar methods in next chapter, like statistically optimized NAH, (SONAH) and iterative NAH with recursive filtration. The main product of the thesis is the practical part represented by testing application. That is created in the Matlab environment and is able to calculate and display hologram of the scanned array by the planar NAH method using the “k-space” filter. The application supposes a planar sound source and in other cases the accuracy of the reconstruction is not guaranteed. There are also given some holograms calculated with the application.