Nowadays, speech synthesis is part of various daily life applications. The ultimate goal of such technologies consists in extending the possibilities of interaction with the machine, in order to get closer to human-like communications. However, current state-of-the-art systems often lack of realism: although high-quality speech synthesis can be produced by many researchers and companies around the world, synthetic voices are generally perceived as hyperarticulated. In any case, their degree of articulation is fixed once and for all. The present thesis falls within the more general quest for enriching expressivity in speech synthesis. The main idea consists in improving statistical parametric speech synthesis, whose most famous example is Hidden Markov Model (HMM) based speech synthesis, by introducing a control of the articulation degree, so as to enable synthesizers to automatically adapt their way of speaking to the contextual situation, like humans ...

Reducing interference noise in a noisy speech recording has been a challenging task for many years yet has a variety of applications, for example, in handsfree mobile communications, in speech recognition, and in hearing aids. Traditional single-channel noise reduction schemes, such as Wiener filtering, do not work satisfactorily in the presence of non-stationary background noise. Alternatively, supervised approaches, where the noise type is known in advance, lead to higher-quality enhanced speech signals. This dissertation proposes supervised and unsupervised single-channel noise reduction algorithms. We consider two classes of methods for this purpose: approaches based on nonnegative matrix factorization (NMF) and methods based on hidden Markov models (HMM). The contributions of this dissertation can be divided into three main (overlapping) parts. First, we propose NMF-based enhancement approaches that use temporal dependencies of the speech signals. In a standard NMF, the important temporal ...

In modern devices which provide hands-free speech capturing functionality, such as hands-free communication kits and voice-controlled devices, the received speech signal at the microphones is corrupted by background noise, interfering speech signals, and room reverberation. In many practical situations, the microphones are not necessarily located near the desired source, and hence, the ratio of the desired speech power to the power of the background noise, the interfering speech, and the reverberation at the microphones can be very low, often around or even below 0 dB. In such situations, the comfort of human-to-human communication, as well as the accuracy of automatic speech recognisers for voice-controlled applications can be signi cantly degraded. Therefore, e ffective speech enhancement algorithms are required to process the microphone signals before transmitting them to the far-end side for communication, or before feeding them into a speech recognition ...

This dissertation's primary goal was to systematically study human activity recognition and enhance its performance by advancing human activities' sequential modeling based on HMM-based machine learning. Driven by these purposes, this dissertation has the following major contributions: The proposal of our HAR research pipeline that guides the building of a robust wearable end-to-end HAR system and the implementation of the recording and recognition software Activity Signal Kit (ASK) according to the pipeline; Collecting several datasets of multimodal biosignals from over 25 subjects using the self-implemented ASK software and implementing an easy mechanism to segment and annotate the data; The comprehensive research on the offline HAR system based on the recorded datasets and the implementation of an end-to-end real-time HAR system; A novel activity modeling method for HAR, which partitions the human activity into a sequence of shared, meaningful, and activity ...

In the early days of speech synthesis research, understanding voice production has attracted the attention of scientists with the goal of producing intelligible speech. Later, the need to produce more natural voices led researchers to use prerecorded voice databases, containing speech units, reassembled by a concatenation algorithm. With the outgrowth of computer capacities, the length of units increased, going from diphones to non-uniform units, in the so-called unit selection framework, using a strategy referred to as 'take the best, modify the least'. Today the new challenge in voice synthesis is the production of expressive speech or singing. The mainstream solution to this problem is based on the “there is no data like more data” paradigm: emotionspecific databases are recorded and emotion-specific units are segmented. In this thesis, we propose to restart the expressive speech synthesis problem, from its original voice ...

Above 20% European citizens speak in Russian therefore the task of automatic recognition of Russian continuous speech has a key significance. The main problems of ASR are connected with the complex mechanism of Russian word-formation. Totally there exist above 3 million diverse valid word-forms that is very large vocabulary ASR task. The thesis presents the novel HMM-based ASR model of Russian that has morphemic levels of speech and language representation. The model includes the developed methods for decomposition of the word vocabulary into morphemes and acoustical and statistical language modelling at the training stage and the method for word synthesis at the last stage of speech decoding. The presented results of application of the ASR model for voice access to the Yellow Pages directory have shown the essential improvement (above 75%) of the real-time factor saving acceptable word recognition rate ...

In natural listening environments, speech signals are easily distorted by variousacoustic interference, which reduces the speech quality and intelligibility of human listening; meanwhile, it makes difficult for many speech-related applications, such as automatic speech recognition (ASR). Thus, many speech enhancement (SE) algorithms have been developed in the past decades. However, most current SE algorithms are difficult to capture underlying speech information (e.g., phoneme) in the SE process. This causes it to be challenging to know what specific information is lost or interfered with in the SE process, which limits the application of enhanced speech. For instance, some SE algorithms aimed to improve human listening usually damage the ASR system. The objective of this dissertation is to develop SE algorithms that have the potential to capture various underlying speech representations (information) and improve the quality and intelligibility of noisy speech. This ...

The topic of this thesis is the enhancement of noisy, periodic signals with application to speech signals. Generally speaking, enhancement methods can be divided into signal- and noise-driven methods. In this thesis, we focus on the signal-driven approach by employing relevant signal parameters for the enhancement of periodic signals. The enhancement problem consists of two major subproblems: the estimation of relevant parameters or statistics, and the actual noise reduction of the observed signal. We consider both of these subproblems. First, we consider the problem of estimating signal parameters relevant to the enhancement of periodic signals. The fundamental frequency is one example of such a parameter. Furthermore, in multichannel scenarios, the direction-of-arrival of the periodic sources onto an array of sensors is another parameter of relevance. We propose methods for the estimation of the fundamental frequency that have benefits compared to ...

This thesis deals with several open problems in acoustic echo cancellation and acoustic feedback control. Our main goal has been to develop solutions that provide a high performance and sound quality, and behave in a robust way in realistic conditions. This can be achieved by departing from the traditional ad-hoc methods, and instead deriving theoretically well-founded solutions, based on results from parameter estimation and system identification. In the development of these solutions, the computational efficiency has permanently been taken into account as a design constraint, in that the complexity increase compared to the state-of-the-art solutions should not exceed 50 % of the original complexity. In the context of acoustic echo cancellation, we have investigated the problems of double-talk robustness, acoustic echo path undermodeling, and poor excitation. The two former problems have been tackled by including adaptive decorrelation filters in the ...

Today, adaptive algorithms represent one of the most frequently used computational tools for the processing of digital speech signals. This work investigates and analyzes the properties of adaptive algorithms in speech communication applications where rigorous conditions apply, such as noise and echo cancelation. Like other theses in this field do, it tries to tackle the ever-lasting problem of computational complexity vs. rate of convergence. It introduces some new adaptive methods that stem from the existing algorithms as well as a novel concept which has been entitled Optimal Step-Size (OSS). In the first part of the thesis we investigate some well-known, widely used adaptive techniques such as the Normalized Least Mean Squares (NLMS) and the Recursive Least Mean Squares (RLS). In spite of the fact that the NLMS and the RLS belong to the "simplest" principles, as far as complexity is ...

Absolute security does not exist: given funding, willpower and the proper technology, every security system can be compromised. However, the objective of the security community should be to develop such applications that the funding, the will, and the resources needed by the attacker to crack the system prevent him from attempting to do so. This Thesis is focused on the vulnerability assessment of biometric systems. Although being relatively young compared to other mature and long-used security technologies, biometrics have emerged in the last decade as a pushing alternative for applications where automatic recognition of people is needed. Certainly, biometrics are very attractive and useful for the final user: forget about PINs and passwords, you are your own key. However, we cannot forget that as any technology aimed to provide a security service, biometric systems are exposed to external attacks which ...

Humans without larynx need to use a substitution voice to re-obtain speech. The electro-larynx (EL) is a widely used device but is known for its unnatural and monotonic speech quality. Previous research tackled these problems, but until now no significant improvements could be reported. The EL speech system is a complex system including hardware (artificial excitation source or sound transducer) and software (control and generation of the artificial excitation signal). It is not enough to consider one separated problem, but all aspects of the EL speech system need to be taken into account. In this thesis we would like to push forward the boundaries of the conventional EL device towards a new bionic electro-larynx speech system. We formulate two overall scenarios: a closed-loop scenario, where EL speech is excited and simultaneously recorded using an EL speech system, and the artificial ...

In a communication system it results undoubtedly of great interest to compress the information generated by the data sources to its most elementary representation, so that the amount of power necessary for reliable communications can be reduced. It is often the case that the redundancy shown by a wide variety of information sources can be modelled by taking into account the probabilistic dependance among consecutive source symbols rather than the probabilistic distribution of a single symbol. These sources are commonly referred to as single or multiterminal sources "with memory" being the memory, in this latter case, the existing temporal correlation among the consecutive symbol vectors generated by the multiterminal source. It is well known that, when the source has memory, the average amount of information per source symbol is given by the entropy rate, which is lower than its entropy ...

Due to significant advances made over the last few decades in the areas of (wireless) networking, communications and microprocessor fabrication, the use of sensor networks to observe physical phenomena is rapidly becoming commonplace. Over this period, many aspects of sensor networks have been explored, yet a thorough understanding of how to analyse and process the vast amounts of sensor data collected remains an open area of research. This work, therefore, aims to provide theoretical, as well as practical, advances this area. In particular, we consider the problem of inferring certain underlying properties of the monitored phenomena, from our sensor measurements. Within mathematics, this is commonly formulated as an inverse problem; whereas in signal processing, it appears as a (multidimensional) sampling and reconstruction problem. Indeed it is well known that inverse problems are notoriously ill-posed and very demanding to solve; meanwhile ...

Many signal and image processing applications have benefited remarkably from the theory of sparse representations. In its classical form this theory models signal as having a sparse representation under a given dictionary -- this is referred to as the "Synthesis Model". In this work we focus on greedy methods for the problem of recovering a signal from a set of deteriorated linear measurements. We consider four different sparsity frameworks that extend the aforementioned synthesis model: (i) The cosparse analysis model; (ii) the signal space paradigm; (iii) the transform domain strategy; and (iv) the sparse Poisson noise model. Our algorithms of interest in the first part of the work are the greedy-like schemes: CoSaMP, subspace pursuit (SP), iterative hard thresholding (IHT) and hard thresholding pursuit (HTP). It has been shown for the synthesis model that these can achieve a stable recovery ...