Deep Learning for i-Vector Speaker and Language Recognition

Over the last few years, i-vectors have been the state-of-the-art technique in speaker and language recognition. Recent advances in Deep Learning (DL) technology have improved the quality of i-vectors but the DL techniques in use are computationally expensive and need speaker or/and phonetic labels for the background data, which are not easily accessible in practice. On the other hand, the lack of speaker-labeled background data makes a big performance gap, in speaker recognition, between two well-known cosine and Probabilistic Linear Discriminant Analysis (PLDA) i-vector scoring techniques. It has recently been a challenge how to fill this gap without speaker labels, which are expensive in practice. Although some unsupervised clustering techniques are proposed to estimate the speaker labels, they cannot accurately estimate the labels. This thesis tries to solve the problems above by using the DL technology in different ways, without ...

Ghahabi, Omid — Universitat Politecnica de Catalunya


Deep Learning for Event Detection, Sequence Labelling and Similarity Estimation in Music Signals

When listening to music, some humans can easily recognize which instruments play at what time or when a new musical segment starts, but cannot describe exactly how they do this. To automatically describe particular aspects of a music piece – be it for an academic interest in emulating human perception, or for practical applications –, we can thus not directly replicate the steps taken by a human. We can, however, exploit that humans can easily annotate examples, and optimize a generic function to reproduce these annotations. In this thesis, I explore solving different music perception tasks with deep learning, a recent branch of machine learning that optimizes functions of many stacked nonlinear operations – referred to as deep neural networks – and promises to obtain better results or require less domain knowledge than more traditional techniques. In particular, I employ ...

Schlüter, Jan — Department of Computational Perception, Johannes Kepler University Linz


Speech derereverberation in noisy environments using time-frequency domain signal models

Reverberation is the sum of reflected sound waves and is present in any conventional room. Speech communication devices such as mobile phones in hands-free mode, tablets, smart TVs, teleconferencing systems, hearing aids, voice-controlled systems, etc. use one or more microphones to pick up the desired speech signals. When the microphones are not in the proximity of the desired source, strong reverberation and noise can degrade the signal quality at the microphones and can impair the intelligibility and the performance of automatic speech recognizers. Therefore, it is a highly demanded task to process the microphone signals such that reverberation and noise are reduced. The process of reducing or removing reverberation from recorded signals is called dereverberation. As dereverberation is usually a completely blind problem, where the only available information are the microphone signals, and as the acoustic scenario can be non-stationary, ...

Braun, Sebastian — Friedrich-Alexander Universität Erlangen-Nürnberg


Deep Learning for Distant Speech Recognition

Deep learning is an emerging technology that is considered one of the most promising directions for reaching higher levels of artificial intelligence. Among the other achievements, building computers that understand speech represents a crucial leap towards intelligent machines. Despite the great efforts of the past decades, however, a natural and robust human-machine speech interaction still appears to be out of reach, especially when users interact with a distant microphone in noisy and reverberant environments. The latter disturbances severely hamper the intelligibility of a speech signal, making Distant Speech Recognition (DSR) one of the major open challenges in the field. This thesis addresses the latter scenario and proposes some novel techniques, architectures, and algorithms to improve the robustness of distant-talking acoustic models. We first elaborate on methodologies for realistic data contamination, with a particular emphasis on DNN training with simulated data. ...

Ravanelli, Mirco — Fondazione Bruno Kessler


Deep neural networks for source separation and noise-robust speech recognition

This thesis addresses the problem of multichannel audio source separation by exploiting deep neural networks (DNNs). We build upon the classical expectation-maximization (EM) based source separation framework employing a multichannel Gaussian model, in which the sources are characterized by their power spectral densities and their source spatial covariance matrices. We explore and optimize the use of DNNs for estimating these spectral and spatial parameters. Employing the estimated source parameters, we then derive a time-varying multichannel Wiener filter for the separation of each source. We extensively study the impact of various design choices for the spectral and spatial DNNs. We consider different cost functions, time-frequency representations, architectures, and training data sizes. Those cost functions notably include a newly proposed task-oriented signal-to-distortion ratio cost function for spectral DNNs. Furthermore, we present a weighted spatial parameter estimation formula, which generalizes the corresponding exact ...

Nugraha, Aditya Arie — Université de Lorraine


Dynamic Scheme Selection in Image Coding

This thesis deals with the coding of images with multiple coding schemes and their dynamic selection. In our society of information highways, electronic communication is taking everyday a bigger place in our lives. The number of transmitted images is also increasing everyday. Therefore, research on image compression is still an active area. However, the current trend is to add several functionalities to the compression scheme such as progressiveness for more comfortable browsing of web-sites or databases. Classical image coding schemes have a rigid structure. They usually process an image as a whole and treat the pixels as a simple signal with no particular characteristics. Second generation schemes use the concept of objects in an image, and introduce a model of the human visual system in the design of the coding scheme. Dynamic coding schemes, as their name tells us, make ...

Fleury, Pascal — Swiss Federal Institute of Technology


Integrating monaural and binaural cues for sound localization and segregation in reverberant environments

The problem of segregating a sound source of interest from an acoustic background has been extensively studied due to applications in hearing prostheses, robust speech/speaker recognition and audio information retrieval. Computational auditory scene analysis (CASA) approaches the segregation problem by utilizing grouping cues involved in the perceptual organization of sound by human listeners. Binaural processing, where input signals resemble those that enter the two ears, is of particular interest in the CASA field. The dominant approach to binaural segregation has been to derive spatially selective filters in order to enhance the signal in a direction of interest. As such, the problems of sound localization and sound segregation are closely tied. While spatial filtering has been widely utilized, substantial performance degradation is incurred in reverberant environments and more fundamentally, segregation cannot be performed without sufficient spatial separation between sources. This dissertation ...

Woodruff, John — The Ohio State University


Mixed structural models for 3D audio in virtual environments

In the world of Information and communications technology (ICT), strategies for innovation and development are increasingly focusing on applications that require spatial representation and real-time interaction with and within 3D-media environments. One of the major challenges that such applications have to address is user-centricity, reflecting e.g. on developing complexity-hiding services so that people can personalize their own delivery of services. In these terms, multimodal interfaces represent a key factor for enabling an inclusive use of new technologies by everyone. In order to achieve this, multimodal realistic models that describe our environment are needed, and in particular models that accurately describe the acoustics of the environment and communication through the auditory modality are required. Examples of currently active research directions and application areas include 3DTV and future internet, 3D visual-sound scene coding, transmission and reconstruction and teleconferencing systems, to name but ...

Geronazzo, Michele — University of Padova


Discrete-time speech processing with application to emotion recognition

The subject of this PhD thesis is the efficient and robust processing and analysis of the audio recordings that are derived from a call center. The thesis is comprised of two parts. The first part is dedicated to dialogue/non-dialogue detection and to speaker segmentation. The systems that are developed are prerequisite for detecting (i) the audio segments that actually contain a dialogue between the system and the call center customer and (ii) the change points between the system and the customer. This way the volume of the audio recordings that need to be processed is significantly reduced, while the system is automated. To detect the presence of a dialogue several systems are developed. This is the first effort found in the international literature that the audio channel is exclusively exploited. Also, it is the first time that the speaker utterance ...

Kotti, Margarita — Aristotle University of Thessaloniki


Transmission over Time- and Frequency-Selective Mobile Wireless Channels

The wireless communication industry has experienced rapid growth in recent years, and digital cellular systems are currently designed to provide high data rates at high terminal speeds. High data rates give rise to intersymbol interference (ISI) due to so-called multipath fading. Such an ISI channel is called frequency selective. On the other hand, due to terminal mobility and/or receiver frequency offset the received signal is subject to frequency shifts (Doppler shifts). Doppler shift induces time-selectivity characteristics. The Doppler effect in conjunction with ISI gives rise to a so-called doubly selective channel (frequency- and time-selective). In addition to the channel effects, the analog front-end may suffer from an imbalance between the I and Q branch amplitudes and phases as well as from carrier frequency offset. These analog front-end imperfections then result in an additional and significant degradation in system performance, especially ...

Barhumi, Imad — Katholieke Universiteit Leuven


Adaptive Digital Predistortion of Nonlinear Systems

Compensating or reducing the nonlinear distortion - usually resulting from a nonlinear system - is becoming an essential requirement in many areas. In this thesis adaptive digital predistortion techniques for a wide class of nonlinear systems are presented. For estimating the coefficients of the predistorter, different learning architectures are considered: the Direct Learning Architecture (DLA) and Indirect Learning Architecture (ILA). In the DLA approach, we propose a new adaptation algorithm - the Nonlinear Filtered-x Prediction Error Method (NFxPEM) algorithm, which has much faster convergence and much better performance compared to the conventional Nonlinear Filtered-x Least Mean Squares (NFxLMS) algorithm. All of these time domain adaptive algorithms require accurate system identification of the nonlinear system. In order to relax or avoid this strict requirement, the NFxLMS with Initial Subsystem Estimates (NFxLMS-ISE) and NFxPEM-ISE algorithms are proposed. Furthermore, we propose a frequency ...

Gan, Li — Graz University of Technology


Analysis and improvement of quantification algorithms for magnetic resonance spectroscopy

Magnetic Resonance Spectroscopy (MRS) is a technique used in fundamental research and in clinical environments. During recent years, clinical application of MRS gained importance, especially as a non-invasive tool for diagnosis and therapy monitoring of brain and prostate tumours. The most important asset of MRS is its ability to determine the concentration of chemical substances non-invasively. To extract relevant signal parameters, MRS data have to be quantified. This usually doesn¢t prove to be straightforward since in vivo MRS signals are characterized by poor signal-to-noise ratios, overlapping peaks, acquisition related artefacts and the presence of disturbing components (e.g. residual water in proton spectra). The work presented in this thesis aims to improve the quantification in different applications of MRS in vivo. To obtain the signal parameters related to MRS data, different approaches were suggested in the past. Black-box methods, don¢t require ...

Pels, Pieter — Katholieke Universiteit Leuven


Continuous-time matrix algorithms systolic algorithms and adaptive neural networks

In the domain of 'continuous-time matrix algorithms', matrix based algorithms are studied from the viewpoint of continuous-time systems theory and differential geometry. We put emphasis on formulas for tracking decompositions of a time-varying matrix, and present them as tools for the design and analysid of matrix algorithms. We define a class of continuous-time matrix algorithms with a uniform parallel signal flow graph. We derive algorithms for recursive least-squares estimation, belonging to this class, which are continuous-time limits of known systolic algorithms. Some of them are candidates for analog realization. For algorithms for subspace tracking, belonging to the same class, we present new analysis results based on continuous-time concepts. From these algorithms we also derive new fully pipelined systolic algorithms, inheriting the main properties of their continuous-time counterparts. We reinterpret the presented continuous-time adaptive signal processing algorithms as adaptation laws for ...

Dehaene, Jeroen — Katholieke Universiteit Leuven


Advanced time-domain methods for nuclear magnetic resonance spectroscopy data analysis

Over the past years magnetic resonance spectroscopy (MRS) has been of significant importance both as a fundamental research technique in different fields, as well as a diagnostic tool in medical environments. With MRS, for example, spectroscopic information, such as the concentrations of chemical substances, can be determined non-invasively. To that end, the signals are first modeled by an appropriate model function and mathematical techniques are subsequently applied to determine the model parameters. In this thesis, signal processing algorithms are developed to quantify in-vivo and ex-vivo MRS signals. These are usually characterized by a poor signal-to-noise ratio, overlapping peaks, deviations from the model function and in some cases the presence of disturbing components (e.g. the residual water in proton spectra). The work presented in this thesis addresses a part of the total effort to provide accurate, efficient and automatic data analysis ...

Vanhamme, Leentje — Katholieke Universiteit Leuven


Robust Direction-of-Arrival estimation and spatial filtering in noisy and reverberant environments

The advent of multi-microphone setups on a plethora of commercial devices in recent years has generated a newfound interest in the development of robust microphone array signal processing methods. These methods are generally used to either estimate parameters associated with acoustic scene or to extract signal(s) of interest. In most practical scenarios, the sources are located in the far-field of a microphone array where the main spatial information of interest is the direction-of-arrival (DOA) of the plane waves originating from the source positions. The focus of this thesis is to incorporate robustness against either lack of or imperfect/erroneous information regarding the DOAs of the sound sources within a microphone array signal processing framework. The DOAs of sound sources is by itself important information, however, it is most often used as a parameter for a subsequent processing method. One of the ...

Chakrabarty, Soumitro — Friedrich-Alexander Universität Erlangen-Nürnberg

The current layout is optimized for mobile phones. Page previews, thumbnails, and full abstracts will remain hidden until the browser window grows in width.

The current layout is optimized for tablet devices. Page previews and some thumbnails will remain hidden until the browser window grows in width.