In communication system environment speech signal is corrupted due to presence of additive acoustic noise, so with this distortion the effective communication is degraded in terms of the quality and intelligibility of speech. Now present research is going how effectively acoustic noise can be eliminated without affecting the original speech quality, this tends to be our challenging in this current research thesis work. Here this work proposes multi-tiered detection method that is based on time-frequency analysis (i.e. filter banks concept) of the noisy speech signals, by using standard speech enhancement method based on the proven spectral subtraction, for single channel speech data and for a wide range of noise types at various noise levels. There were various variants have been introduced to standard spectral subtraction proposed by S.F.Boll. In this thesis we designed and implemented a novel approach of Spectral Subtraction based on Minimum Statistics [MinSSS]. This means that the power spectrum of the non-stationary noise signal is estimated by finding the minimum values of a smoothed power spectrum of the noisy speech signal and thus circumvents the speech activity detection problem. This approach is also capable of dealing with non-stationary noise signals. In order to analyze the system in time frequency domain, we have implemented two different filter bank approaches such as Weighted OverLap Added (WOLA) and Fast Fourier Transform Modulated (FFTMod). The proposed systems were implemented and evaluated offline using simulation tool Matlab and then validated their performances based on the objective quality measures such as Signal to Noise Ratio Improvement (SNRI) and Perceptual Evaluation Speech Quality (PESQ) measure. The systems were tested with a pure speech combination of male and female sampled at 8 kHz, these signals were corrupted with various kinds of noises at different noise power levels. The MinSSS algorithm implemented using FFTMod filter bank approach outperforms when compared the WOLA filter bank approach.

Orientador: Jozué Vieira Filho / Banca: Suely Cunha Amaro Mantovani / Banca: Marco Aparecido Queiroz Duarte / Resumo: A proposta deste trabalho é aprimorar a performance da técnica de redução de ruído, subtração espectral baseado na relação SNR a Priori, através da implementação de dois novos parâmetros Potência de Articulação e Não-Articulação obtidas a partir de algumas técnicas psicoacústicas. Faz-se um estudo da anatomia do sistema de audição humana e algumas limitações físicas, com o objetivo de entender o princípio básico da técnica ANIQUE, que é um sistema de avaliação objetiva de voz e têm como princípio o modelamento da percepção humana da voz. Através do modelo ANIQUE são extraídas as principais técnicas psicoacústicas para obtenção dos novos parâmetros, Potência de Articulação e Não- Articulação. Procurou-se apresentar de maneira resumida o processo de equacionamento das técnicas de redução de ruído em sinais de voz e das técnicas psicoacústicas. Posteriormente são descritos todos os processos das técnicas utilizadas que foram simuladas utilizando a linguagem de programação do MatLab®, seguido das avaliações objetivas dos sinais processados pelo software PESQ, que é um programa de avaliação objetiva de voz. Os resultados mostram que a implementação das técnicas psicoacústicas foram eficazes para melhorar a performance da técnica subtração espectral baseada na relação SNR a Priori / Abstract: The purpose of this work is to enhance the performance of noise reduction techniques based on spectral subtraction, which take in account the a priori signal-to-noise (SNR a Priori) to be estimated considering psychoacoustic criteria. in order to understand the basic principle of the ANIQUE, which is a psychoacoustic based technique used to evaluate the quality of speech signals, it was necessary to develop a study of the anatomy of the human hearing and some physical limitations, From the ANIQUE are extracted new parameters namely Articulation and Non-Articulation Powers, used to estimate the SNR_prio. As a result, it was obtained a new spectral based technique which was implemented in the MatLab® environment and evaluated using the objective quality measure for speech signal simulations namely PESQ. The results show that the implementation of psychoacoustic techniques were effective in enhance the performance of the spectral subtraction technique based on SNR a Priori / Mestre

Made available in DSpace on 2014-06-11T19:29:49Z (GMT). No. of bitstreams: 0 Previous issue date: 2010-02-25Bitstream added on 2014-06-13T19:59:46Z : No. of bitstreams: 1 kanda_az_me_ilha.pdf: 1642888 bytes, checksum: 49ebcd86d5690e7be50fcdca35a52a48 (MD5) / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) / A proposta deste trabalho é aprimorar a performance da técnica de redução de ruído, subtração espectral baseado na relação SNR a Priori, através da implementação de dois novos parâmetros Potência de Articulação e Não-Articulação obtidas a partir de algumas técnicas psicoacústicas. Faz-se um estudo da anatomia do sistema de audição humana e algumas limitações físicas, com o objetivo de entender o princípio básico da técnica ANIQUE, que é um sistema de avaliação objetiva de voz e têm como princípio o modelamento da percepção humana da voz. Através do modelo ANIQUE são extraídas as principais técnicas psicoacústicas para obtenção dos novos parâmetros, Potência de Articulação e Não- Articulação. Procurou-se apresentar de maneira resumida o processo de equacionamento das técnicas de redução de ruído em sinais de voz e das técnicas psicoacústicas. Posteriormente são descritos todos os processos das técnicas utilizadas que foram simuladas utilizando a linguagem de programação do MatLab®, seguido das avaliações objetivas dos sinais processados pelo software PESQ, que é um programa de avaliação objetiva de voz. Os resultados mostram que a implementação das técnicas psicoacústicas foram eficazes para melhorar a performance da técnica subtração espectral baseada na relação SNR a Priori / The purpose of this work is to enhance the performance of noise reduction techniques based on spectral subtraction, which take in account the a priori signal-to-noise (SNR a Priori) to be estimated considering psychoacoustic criteria. in order to understand the basic principle of the ANIQUE, which is a psychoacoustic based technique used to evaluate the quality of speech signals, it was necessary to develop a study of the anatomy of the human hearing and some physical limitations, From the ANIQUE are extracted new parameters namely Articulation and Non-Articulation Powers, used to estimate the SNR_prio. As a result, it was obtained a new spectral based technique which was implemented in the MatLab® environment and evaluated using the objective quality measure for speech signal simulations namely PESQ. The results show that the implementation of psychoacoustic techniques were effective in enhance the performance of the spectral subtraction technique based on SNR a Priori

This diploma thesis deals with methods of noise suppression for speech recognition systems. In theoretical part are discussed basic terms of this topic and also methods for noise suppression. These are spectral subtraction, Wiener filtering, RASTA, mapping of spectrogram or algorithms based on noise estimation. In second part types of noise are analyzed, there is proposal and implementation of spectral subtraction method of noise suppression for speech recognition system. Also extensive testing of spectral subtractive algorithms in comparison with Wiener filter is conducted. Assessment of this testing is done with defined metrics, successfulness of recognition, recognition system score and signal to noise ratio.

The thesis deals with single channel and multiple channel algorithms for speech enhancement. The goal of this work is to perform the deep analysis of both single channel and multiple channel algorithms in sense of their behaviour in noisy environment of combat vehicles and platforms. Based on this analysis a new composite speech enhancement algorithm will be designed. This new approach is expected to increase quality of the processed speech in military communications systems. These systems are characterised by their operation under very noisy conditions where background noise is very high and signal-to-noise ratio extremely low. These noisy conditions are typical for the range of military and combat platforms and vehicles.

L’objectif de cette thèse est de contribuer à l’amélioration du processus de dépouillement d’essais de sûreté visant étudier le comportement d'un combustible nucléaire en contexte d’accident d’injection de réactivité (RIA), via la technique de contrôle par émission acoustique. Il s’agit notamment d’identifier clairement les mécanismes physiques pouvant intervenir au cours des essais à travers leur signature acoustique. Dans un premier temps, au travers de calculs analytiques et des simulation numériques conduites au moyen d’une méthode d’éléments finis spectraux, l’impact du dispositif d’essais sur la propagation des ondes est étudié. Une fréquence de résonance du dispositif est identifiée. On établit également que les mécanismes basses fréquences ne sont pas impactés par le dispositif d'essais. En second lieu, diverses techniques de traitement du signal (soustraction spectrale, analyse spectrale singulière, ondelettes. . . ) sont expérimentées, afin de proposer des outils permettant de traiter différent types de bruit survenant lors des essais RIA. La soustraction spectrale s’avère être la méthode la plus robuste aux changements de nature du bruit, avec un fort potentiel d’amélioration du rapport signal-à-bruit. Enfin, des méthodes d’analyse de données multivariées et d’analyse de données fonctionnelles ont été appliquées, afin de proposer un algorithme de classification statistique permettant de mieux comprendre la phénoménologie des accidents de type RIA et d’identifier les mécanismes physiques. Selon l’approche (multivariée ou fonctionnelle), les algorithmes obtenus permettent de reconnaître le mécanisme associé à une salve dans plus de 80% des cas. / The objective of the thesis is to contribute to the improvement of the monitoring process of nuclear safety experiments dedicated to study the behavior of the nuclear fuel in a reactivity initiated accident (RIA) context, by using the acoustic emission technique. In particular, we want to identify the physical mechanisms occurring during the experiments through their acoustic signatures. Firstly, analytical derivations and numerical simulations using the spectral finite element method have been performed in order to evaluate the impact of the wave travelpath in the test device on the recorded signals. A resonant frequency has been identified and it has been shown that the geometry and the configuration of the test device may not influence the wave propagation in the low frequency range. Secondly, signal processing methods (spectral subtraction, singular spectrum analysis, wavelets,…) have been explored in order to propose different denoising strategies according to the type of noise observed during the experiments. If we consider only the global SNR improvement ratio, the spectral subtraction method is the most robust to changes in the stochastic behavior of noise. Finally, classical multivariate and functional data analysis tools are used in order to create a machine learning algorithm dedicated to contribute to a better understanding of the phenomenology of RIA accidents. According to the method (multivariate or functional), the obtained algorithms allow to identify the mechanisms in more than 80 % of cases.

Centers for Disease Control and Prevention (CDC) estimate that more than 11.2 million people require critical and emergency care in the United States per year. Optimizing and improving patient morbidity and mortality outcomes are the primary objectives of monitoring in critical and emergency care. Patients in need of critical or emergency care in general are at a risk of single or multiple organ failures occurring due to a traumatic injury, a surgical event, or an underlying pathology that results in severe patient hemodynamic instability. Hence, continuous monitoring of fundamental cardiovascular hemodynamic parameters, such as heart rate, respiratory rate, blood pressure, blood oxygenation and core temperature, is essential to accomplish diagnostics in critical and emergency care. Today’s standard of care measures these critical parameters using multiple monitoring technologies. Though it is possible to measure all the fundamental cardiovascular hemodynamic parameters using the blood flow dynamics, its use is currently only limited to measuring continuous blood pressure. No other comparable studies in the literature were successful in quantifying other critical parameters from the blood flow dynamics for a few reasons. First, the blood flow dynamics exhibit a complicated and sensitive dynamic pressure field. Existing blood flow based data acquisition systems are unable to detect these sensitive variations in the pressure field. Further, the pressure field is also influenced by the presence of background acoustic interference, resulting in a noisy pressure profile. Thus in order to extract critical parameters from this dynamic pressure field with fidelity, there is need for an integrated framework that is composed of a highly sensitive data acquisition system and advanced signal processing. In addition, existing state-of-the-art technologies require expensive instrumentation and complex infrastructure. The information sensed using these multiple monitoring technologies is integrated and visualized using a clinical information system. This process of integration and visualization creates the need for functional interoperability within the multiple monitoring technologies. Limited functional interoperability not only results in diagnostic errors but also their complexity makes it impossible to use such technologies to accomplish monitoring in low resource settings. These multiple monitoring technologies are neither portable nor scalable, in addition to inducing extreme patient discomfort. For these reasons, existing monitoring technologies do not efficiently meet the monitoring and diagnostic requirements of critical and emergency care. In order to address the challenges presented by existing blood flow based data acquisition systems and other monitoring systems, a point of care monitoring device was developed to provide multiple critical parameters by means of uniquely measuring a physiological process. To demonstrate the usability of this novel catheter multiscope, a feasibility study was performed using an animal model. The corresponding results are presented in this dissertation. The developed measurement system first acquires the dynamics of blood flow through a minimally invasive catheter. Then, a signal processing framework is developed to characterize the blood flow dynamics and to provide critical parameters such as heart rate, respiratory rate, and blood pressure. The framework used to extract the physiological data corresponding to the acoustic field of the blood flow consisted of a noise cancellation technique and a wavelet based source separation. The preliminary results of the acoustic field of the blood flow revealed the presence of acoustic heart and respiratory pulses. A unique and novel framework was also developed to extract continuous blood pressure from the pressure field of the blood flow. Finally, the computed heart and respiratory rates, systolic and diastolic pressures were benchmarked with actual values measured using conventional devices to validate the measurements of the catheter multiscope. In summary, the results of the feasibility study showed that the novel catheter multiscope can provide critical parameters such as heart rate, respiratory rate and blood pressure with clinical accuracy. In addition, this dissertation also highlights the diagnostic potential of the developed catheter multiscope by presenting preliminary results of proof of concept studies performed for application case studies such as sinus rhythm pattern recognition and fetal monitoring through phonocardiography.

In speech communication systems such as voice-controlled systems, hands-free mobile telephones and hearing aids, the received signals are degraded by room reverberation and background noise. This degradation can reduce the perceived quality and intelligibility of the speech, and decrease the performance of speech enhancement and source localization. These problems are difficult to solve due to the colored and nonstationary nature of the speech signals, and features of the Room Impulse Response (RIR) such as its long duration and non-minimum phase. In this dissertation, we focus on two topics of speech enhancement and speaker localization in noisy reverberant environments. A two-stage speech enhancement method is presented to suppress both early and late reverberation in noisy speech using only one microphone. It is shown that this method works well even in highly reverberant rooms. Experiments under different acoustic conditions confirm that the proposed blind method is superior in terms of reducing early and late reverberation effects and noise compared to other well known single-microphone techniques in the literature. Time Difference Of Arrival (TDOA)-based methods usually provide the most accurate source localization in adverse conditions. The key issue for these methods is to accurately estimate the TDOA using the smallest number of microphones. Two robust Time Delay Estimation (TDE) methods are proposed which use the information from only two microphones. One method is based on adaptive inverse filtering which provides superior performance even in highly reverberant and moderately noisy conditions. It also has negligible failure estimation which makes it a reliable method in realistic environments. This method has high computational complexity due to the estimation in the first stage for the first microphone. As a result, it can not be applied in time-varying environments and real-time applications. Our second method improves this problem by introducing two effective preprocessing stages for the conventional Cross Correlation (CC)-based methods. The results obtained in different noisy reverberant conditions including a real and time-varying environment demonstrate that the proposed methods are superior compared to the conventional TDE methods. / Graduate / 0544 / 0984 / saeed.mosayyebpour@gmail.com

In speech communication systems such as voice-controlled systems, hands-free mobile telephones and hearing aids, the received signals are degraded by room reverberation and background noise. This degradation can reduce the perceived quality and intelligibility of the speech, and decrease the performance of speech enhancement and source localization. These problems are difficult to solve due to the colored and nonstationary nature of the speech signals, and features of the Room Impulse Response (RIR) such as its long duration and non-minimum phase. In this dissertation, we focus on two topics of speech enhancement and speaker localization in noisy reverberant environments. A two-stage speech enhancement method is presented to suppress both early and late reverberation in noisy speech using only one microphone. It is shown that this method works well even in highly reverberant rooms. Experiments under different acoustic conditions confirm that the proposed blind method is superior in terms of reducing early and late reverberation effects and noise compared to other well known single-microphone techniques in the literature. Time Difference Of Arrival (TDOA)-based methods usually provide the most accurate source localization in adverse conditions. The key issue for these methods is to accurately estimate the TDOA using the smallest number of microphones. Two robust Time Delay Estimation (TDE) methods are proposed which use the information from only two microphones. One method is based on adaptive inverse filtering which provides superior performance even in highly reverberant and moderately noisy conditions. It also has negligible failure estimation which makes it a reliable method in realistic environments. This method has high computational complexity due to the estimation in the first stage for the first microphone. As a result, it can not be applied in time-varying environments and real-time applications. Our second method improves this problem by introducing two effective preprocessing stages for the conventional Cross Correlation (CC)-based methods. The results obtained in different noisy reverberant conditions including a real and time-varying environment demonstrate that the proposed methods are superior compared to the conventional TDE methods. / Graduate / 2015-04-23 / 0544 / 0984 / saeed.mosayyebpour@gmail.com

[pt] Este trabalho apresenta um estudo comparativo de três técnicas de melhoria das taxas de reconhecimento de voz em ambiente adverso, a saber: Normalização da Média Cepestral (CMN), Subtração Espectral e Regressão Linear no Sentido da Máxima Verossimilhança (MLLR), aplicadas isoladamente e em concomitância, duas a duas. Os testes são realizados usando um sistema simples: reconhecimento de palavras isoladas (dígitos de zero a nove, e meia), modo dependente do locutor, modelos ocultos de Markov do tipo contínuo, e vetores de atributos com doze coeficientes cepestrais derivados da análise de predição linear. São adotados três tipos de ruído (gaussiano branco, falatório e de fábrica) em nove razões sinal-ruído diferentes. Os resultados experimentais demonstram que o emprego isolado das técnicas de reconhecimento robusto é, em geral, vantajoso, pois nas diversas razões sinal-ruído para as quais os testes são efetuados, quando as taxas de reconhecimento não sofrem um acréscimo, mantém-se as mesmas obtidas quando não se aplica nenhum método de aumento da robustez. Analisando-se comparativamente as implementações isoladas e simultânea das técnicas, constata-se que a simultânea nem sempre é atraente, dependendo da dupla empregada. Apresentam-se, ainda, os resultados decorrentes do uso de modelos ruidosos, observando-se que, embora sejam inegavelmente melhores, sua utilização é inviável na prática. Das técnicas implementadas, a que representa resultados mais próximos ao emprego de modelos ruidosos é a MLLR, seguida pela CMN, e por último pela Subtração Espectral. Estas últimas, embora percam em desempenho para a primeira, apresentam como vantagem a simplicidade e a generalidade. No que concerne as técnicas usadas concomitantemente, a dupla Subtração Espectral e MLLR é a considerada de melhor performance, pois mostra-se conveniente em relação ao emprego isolado de ambos os métodos, o que nem sempre ocorre com o uso de outras combinações das técnicas individuais. / [en] This work presents a comparative study of three techniques for improving the speech recognition rates in adverse environment, namely: Cepstral Mean Normalization (CMN), Spectral Subtraction and Maximum Likelihood Linear Regression (MLLR). They are implemented in two ways: separately and in pairs. The tests are carried out on a simple system: recognition of isolated words (digits from zero to nine, and the word half), speaker-dependent mode, continuous hidden Markov models, and speech feature vectors with twelve cepstral coefficients derived from linear predictive analysis. Three types of noise are considered (the white one, voice babble and from factory) at nine different signal-to-noise ratios. Experimental result demonstrate that it is worth using separately the techniques of robust recognition. This is because for all signal-to-noise conditions when the recognition accuracy is not improved it is the same one obtained when no method for increasing the robustness is applied. Analyzing comparatively the isolated and simultaneous applications of the techniques, it is verified that the later is not always more attractive than the former one. This depends on the pair of techniques. The use of noisy models is also considered. Although it presents better results, it is not feasible to implement in pratical situations. Among the implemented techniques, MLLR presents closer results to the ones obtaneid with noisy models, followed by CMN, and, at last, by Spectral Subtraction. Although the two later ones are beaten by the first, in terms of recognition accuracy, their advantages are the simplicity and the generality. The use of simultaneous techniques reveals that the pair Spectral Subtraction and MLLR is the one with the best performance because it is superior in comparison with the individual use of both methods. This does not happen with other combination of techniques. / [es] Este trabajo presenta un estudio comparativo de tres técnicas de mejoría de las tasas de reconocimiento de voz en ambiente adverso, a saber: Normalización de la Media Cepextral (CMN), Substracción Espectral y Regresión Lineal en el Sentido de la Máxima Verosimilitud (MLLR), aplicadas separada y conjuntamente, dos a dos. Las pruebas son realizados usando un sistema simple: reconocimiento de palabras aisladas (dígitos de cero al nueve, y media), de modo dependiente del locutor, modelos ocultos de Markov de tipo contínuo, y vectores de atributos con doce coeficientes cepextrales derivados del análisis de predicción lineal. Se adoptan tres tipos de ruido (gausiano blanco, parlatorio y de fábrica) en nueve razones señal- ruido diferentes. Los resultados experimentales demuestran que el empleo aislado de las técnicas de reconocimiento robusto es, en general, ventajoso, pues en las diversas relaciones señal ruido para las cuales las pruebas son efetuadas, cuando la tasa de reconocimiento no aumenta, manteniendo las mismas tasas cuando no se aplica ningún método de aumento de robustez. Analizando comparativamente las implementaciones aisladas y simultáneas de las técnicas, se constata que no siempre la simultánea resulta atractiva, dependiendo de la dupla utilizada. Se presentan además los resultados al utilizar modelos ruidosos, observando que, aunque resultan mejores, su utilización em la práctica resulta inviable. De las técnicas implementadas, la que presenta resultados más próximos al empleo de modelos ruidosos es la MLLR, seguida por la CMN, y por último por la Substracción Espectral. Estas últimas, aunque tienen desempeño peor que la primera, tienen como ventaja la simplicidad y la generalidad. En lo que se refiere a las técnicas usadas concomitantemente, la dupla Substracción Espectral y MLLR es la de mejor performance, pues se muestra conveniente en relación al empleo aislado de ambos métodos, lo que no siempre ocurre con el uso de otras combinaciones de las técnicas individuales.