De nos jours, le maintien à domicile des personnes éprouvant des troubles cognitifs est un réel défi. En effet, les centres spécialisés pour ces personnes ne sont pas suffisants et manquent parfois de ressources. C'est là que l'assistance à domicile intervient pour permettre à ces personnes de conserver un certain niveau d'autonomie avec le moins d'intrusion possible dans leur vie quotidienne. Pour répondre à cette demande, le laboratoire DOMUS s'est équipé d'un appartement intelligent pilote muni des Nouvelles Technologies de l'Information et de Communication (NTIC). L'infrastructure du laboratoire est composée entre autres de capteurs infrarouges, de capteurs de pression, d'écrans tactiles et de microphones qui ont pour but de rendre compte des activités de la vie quotidienne (AVQ) réalisées par la personne au sein de l'habitat. La présente étude rapporte les résultats d'une partie du système d'assistance à domicile permettant l'identification des AVQ par une personne au sein d'un appartement intelligent, à l'aide d'un unique microphone par pièce. Dès lors, on s'intéresse à la détection et la reconnaissance des événements sonores générés par la réalisation des AVQ de l'habitant. Les événements sonores testés correspondent aux activités qui ont lieu dans une cuisine, telle que des bruits de cuisson, d'écoulement d'eau, d'ustensile, de casserole, de vaisselle et de claquements de porte.

No description available.

published_or_final_version / Speech and Hearing Sciences / Master / Master of Science in Audiology

The complex relationship between array gain patterns and microphone distributions limits the application of traditional optimization algorithms on irregular arrays, which show enhanced beamforming performance for human speech capture in immersive environments. This work analyzes the relationship between irregular microphone geometries and spatial filtering performance with statistical methods. Novel geometry descriptors are developed to capture the properties of irregular microphone distributions showing their impact on array performance. General guidelines and optimization methods for regular and irregular array design are proposed in immersive (near-field) environments to obtain superior beamforming ability for speech applications. Optimization times are greatly reduced through the objective function rules using performance-based geometric descriptions of microphone distributions that circumvent direct array gain computations over the space of interest. In addition, probabilistic descriptions of acoustic scenes are introduced to incorporate various levels of prior knowledge for the source distribution. To verify the effectiveness of the proposed optimization methods, simulated gain patterns and real SNR results of the optimized arrays are compared to corresponding traditional regular arrays and arrays obtained from direct exhaustive searching methods. Results show large SNR enhancements for the optimized arrays over arbitrary randomly generated arrays and regular arrays, especially at low microphone densities. The rapid convergence and acceptable processing times observed during the experiments establish the feasibility of proposed optimization methods for array geometry design in immersive environments where rapid deployment is required with limited knowledge of the acoustic scene, such as in mobile platforms and audio surveillance applications.

Avaliou-se a eficácia de um procedimento de teleconsulta baseado na internet para a verificação do aparelho de amplificação sonora individual (AASI) por meio da realização de medidas com microfone sonda em adultos deficientes auditivos. Foram comparados os valores da amplitude da resposta de ressonância da orelha externa (REUR), resposta de ressonância da orelha externa com uso da amplificação (REAR) e o ganho de inserção (REIG) realizados face a face (F) e à distância (D). O erro casual das medidas da REUR, REAR e REIG realizadas pelos dois métodos foi analisado. Participaram do estudo: grupo A: sessenta adultos (média de idade de 67 anos) com deficiência auditiva unilateral (n=15) ou bilateral (n=45), totalizando 105 orelhas; grupo B: 19 adultos (média de idade de 28 anos) sem queixas auditivas (19 orelhas). Os AASIs utilizados foram do tipo mini-retroauricular e não possuíam microfone direcional ou algoritmos de cancelamento de retroalimentação e/ou redução de ruído. Para o grupo B foi feito ajuste do AASI resultando na menor saída possível. Cinco fonoaudiólogos voluntários auxiliaram na coleta dos dados à distância. O procedimento face a face foi realizado pela pesquisadora em todos os participantes, utilizando-se o equipamento Unity PC Probe Mic (Siemens) conectado a um computador pessoal. Para realização das medidas à distância esse computador foi conectado à LAN (10 Mpbs). O participante e o fonoaudiólogo voluntário se posicionaram na sala junto ao equipamento (ambiente de teste). A pesquisadora posicionou-se em outra sala denominada ambiente remoto juntamente com um notebook também conectado à LAN. Webcams e headsets foram utilizados para a captura de audio e video, transmitidas em tempo real pelo software Polycom PVX, o qual também foi utilizado para compartilhamento de aplicativos. Por meio desse compartilhamento a pesquisadora realizava as medidas diretamente no participante, com auxílio do fonoaudiólogo voluntário. Quatro repetições das medidas da REUR, REAR e REIG foram obtidas apenas para o grupo B. Correlações (Pearson) muito fortes e significativas foram obtidas entre as medidas realizadas face a face (F) e à distância (D). O teste t pareado revelou diferenças pequenas, porém significativas entre as medidas F e D para a REUR e REAR. A ANOVA mostrou diferenças significativas entre as repetições realizadas. Diferenças entre os métodos F e D foram obtidas apenas na freqüência de 2000 Hz. Os erros causais obtidos pelas medidas face a face e à distância foram muito semelhantes. As diferenças e variações encontradas entre as medidas face a face e à distância não foram maiores do que a magnitude de variabilidade do próprio procedimento de medidas com microfone sonda. É possível realizar medidas de microfone sonda confiáveis via teleconsulta / telessaúde baseada na internet. Outras investigações são necessárias para validar esse procedimento. / The efficacy of an internet based teleconsultation for hearing aid probe microphone measures in hearing impaired adults was evaluated. The amplitudes of face to face (F) and remote (R) real ear unaided response (REUR), real ear aided response (REAR) and real ear insertion gain (REIG) were compared. The measurement error for repeated measures of REUR, REAR and REIG was analyzed. Participated in this study: group A: 60 adults (mean age: 67 years), with unilateral (n=15) or bilateral (n=45) hearing loss, totalizing 105 ears; group B: 19 normal hearing adults (mean age: 28 years) totalizing 19 ears. Behind the ear hearing aids with no directional microphone, noise reduction and/or feedback cancellation were used. For group B these hearing aids were adjusted as to produce the least amount of gain and output. Five audiologists volunteers helped the data collection. The Unity PC Probe Mic (Siemens) coupled to a personal computer were used by the researcher to carry out face to face measures. For the remote measures this equipment was connected to a local area network (LAN 10 Mpbs). The volunteer and the participant were located in this room (test site). At the remote site the researcher used a notebook connected to the LAN. Webcams and headsets were used for audio and video capture which was transmitted in real time by the Polycom PVX software, which was also used for application sharing. By means of remote controlling of the PC Probe Mic equipment the researcher could perform the remote real ear measurements in the participants. Four repeated measures of REUR, REAR and REIG were obtained for group B only. Strong and significant correlations (Pearson) were obtained between face to face and remote real ear measures. Paired t tests revealed small but significant differences between face to face and remote REUR and REAR. Analysis of variance showed significant differences between repeated measures. Measurement errors for face to face and remote real ear measures were very similar. The differences as well as errors found between face to face and remote measures were never higher than the reported variability for probe microphone measures themselves. It is possible to carry out probe microphone measures by means of teleconsultation / telehealth. Further investigations to validate this procedure are necessary.

<div>Fluid power is widely used in a variety of applications such as construction machines, aerospace, automotive, agricultural machinery, manufacturing, etc. Although this technology has many obvious advantages such as compactness, robustness, high power density, and so forth, there is much room for improvement, of which one of the most important and challenging problems is the noise.</div><div><br></div><div>Different institutes have been researching fluid power noise for decades. However, much of the experimental investigation was based on simple measurement and analysis techniques, which left the designers/researchers no method of understanding the complicated phenomena. A microphone array is a powerful tool that unfortunately has not been introduced to the fluid power noise research. By capturing the magnitude and phase information in space, a microphone array enables the noise source identification, separation, localization and so forth.</div><div><br></div><div>This thesis focuses on implementing the microphone array processing techniques on a synthetic microphone array for fluid power noise diagnostics. Differing from traditional scan-based approaches, the synthetic array is created by synchronizing the non-synchronous measurements to achieve the equivalent effect of a multi-microphone snapshot. The final results will show the power of microphone arrays and provide an economical solution to achieve approximate results when a real microphone array is not available.</div>

"Speech enhancement in an automobile is a challenging problem because interference can come from engine noise, fans, music, wind, road noise, reverberation, echo, and passengers engaging in other conversations. Hands-free microphones make the situation worse because the strength of the desired speech signal reduces with increased distance between the microphone and talker. Automobile safety is improved when the driver can use a hands-free interface to phones and other devices instead of taking his eyes off the road. The demand for high quality hands-free communication in the automobile requires the introduction of more powerful algorithms. This thesis shows that a unique combination of five algorithms can achieve superior speech enhancement for a hands-free system when compared to beamforming or spectral subtraction alone. Several different designs were analyzed and tested before converging on the configuration that achieved the best results. Beamforming, voice activity detection, spectral subtraction, perceptual nonlinear weighting, and talker isolation via pitch tracking all work together in a complementary iterative manner to create a speech enhancement system capable of significantly enhancing real world speech signals. The following conclusions are supported by the simulation results using data recorded in a car and are in strong agreement with theory. Adaptive beamforming, like the Generalized Side-lobe Canceller (GSC), can be effectively used if the filters only adapt during silent data frames because too much of the desired speech is cancelled otherwise. Spectral subtraction removes stationary noise while perceptual weighting prevents the introduction of offensive audible noise artifacts. Talker isolation via pitch tracking can perform better when used after beamforming and spectral subtraction because of the higher accuracy obtained after initial noise removal. Iterating the algorithm once increases the accuracy of the Voice Activity Detection (VAD), which improves the overall performance of the algorithm. Placing the microphone(s) on the ceiling above the head and slightly forward of the desired talker appears to be the best location in an automobile based on the experiments performed in this thesis. Objective speech quality measures show that the algorithm removes a majority of the stationary noise in a hands-free environment of an automobile with relatively minimal speech distortion."

Thesis (Ocean E)--Massachusetts Institute of Technology, Dept. of Ocean Engineering, 1982. / MICROFICHE COPY AVAILABLE IN ARCHIVES AND ENGINEERING / Includes bibliographical references. / by Carl Alzen Wales. / Ocean E

The geometrical characteristics of acoustical radiation are of great importance in instrument design and synthesis, and multiple simplified models have been developed in the past to describe them. In this work two experimental methodologies are proposed and carried out, studying the frequency-dependent radiation in a collection of popular brass instruments with different grades of flaring, and making use of the axis-symmetry of these instruments. The first method uses a scanning linear array and is carefully designed to extract the linear properties of the radiation field. The results of this experimental method are a database of impulse responses distributed in space, and effectively covering a bidimensional on-axis section of the radiation field approximately 0.6 m by 0.9 m. These data can then be used for the validation of a number of simplified physical models used to describe the radiation of these types of instruments. The second method aims at visualising radiation for high amplitude excitation, where shock waves are generated inside the instrument due to non-linear propagation of the plane wave. In this case, the experimental methodology used, taking advantage of the strong density and temperature gradients generated in the air, is an on-axis schlieren optical system. General results of this visualisation show a strong increase in focused directivity at high frequencies and loud playing dynamics, due to the spectral enrichment typical of this family of instruments. The second section of this thesis focuses on the study of playing gestures in the trombone, and could also be applicable to other slide instruments. During glissando playing in the trombone the length of the cylindrical slide section within the bore is altered while waves are propagating. Slide velocities of 2 metres per second are not unusual and result in a (small but measurable) Doppler shift in the wave coming from the mouthpiece before it arrives at the bell. An additional effect is observed in terms of the volume of air within the instrument changing telescopically, leading to a localised change in DC pressure and a resulting flow, which generates infrasound components within the bore. The effects of these playing gestures are investigated in two different setups; one with a high frequency sinusoidal excitation generated by a compression driver, and another one using an artificial mouth to play the instrument. In both experiments the pressures at the mouth or mouthpiece, water key and bell were tracked using microphones and the position of the slide was tracked using a laser distance sensor. Both Doppler shifting and infrasound components were detected for both experimental setups, although the effect on a soft termination such as the artificial lips requires further examination.

Automatic speech recognition has become a standard feature on many consumer electronics and automotive products, and the accuracy of the decoded speech has improved dramatically over time. Often, designers of these products achieve accuracy by employing microphone arrays and beamforming algorithms to reduce interference. However, beamforming microphone arrays are too large for small form factor products such as smart watches. Yet these small form factor products, which have precious little space for tactile user input (i.e. knobs, buttons and touch screens), would benefit immensely from a user interface based on reliably accurate automatic speech recognition. This thesis proposes a solution for interference mitigation that employs blind source separation with a compact array of commercially available unidirectional microphone elements. Such an array provides adequate spatial diversity to enable blind source separation and would easily fit in a smart watch or similar small form factor product. The solution is characterized using publicly available speech audio clips recorded for the purpose of testing automatic speech recognition algorithms. The proposal is modelled in different interference environments and the efficacy of the solution is evaluated. Factors affecting the performance of the solution are identified and their influence quantified. An expectation is presented for the quality of separation as well as the resulting improvement in word error rate that can be achieved from decoding the separated speech estimate versus the mixture obtained from a single unidirectional microphone element. Finally, directions for future work are proposed, which have the potential to improve the performance of the solution thereby making it a commercially viable product.