Source Modeling

This paper describes the use of multi-way decomposition methods to efficiently summarize electroencephalographic (EEG) data. A space-frequency-time atomic decomposition was applied to EEG data recorded while subjects performed tasks associated with varying levels of cognitive performance. The new atomic decomposition of cognitive workload data revealed alpha and theta EEG oscillations which agree with observations reported in the brain research literature. The temporal signature of the atoms discriminates between different levels of cognitive activity. The results and analysis confirm the utility of the multi-way decomposition method to construct new models and algorithms for monitoring cognitive status, which can supplement or overcome existing approaches based on conventional two-dimensional space-time or frequency-time data decomposition.

In this study, we propose an unsupervised classification scheme based on the Dempster-Shafer Theory (TDS) and the Dezert-Smarandache Theory (DSmT) to characterize vegetated, aquatic and mineral surfaces. From pre-processed ASTER satellite images (georeferencing, geometric correction and 15 m re-sampling), neo-channels were produced by determining the spectral indices NDVI, MNDWI and NDBaI, considered as sources of information for classification of a given pixel. NDVI is a contrast function to highlight vegetation. By account, the MNDWI makes it possible to characterize the water and the NDBaI makes it possible to recognize the mineral resources. Then, we modeled respectively the formalisms of the DST and the DSmT, these formalisms are modeling tools close to advanced probabilities based on the notions of belief and fusion functions to take into account certain imperfections (uncertainty, ignorance, etc.) encountered in the acquisition of images. In addition, the DST manages a formalism of disjunction between the sources during the DSmT simultaneously manages a disjunction and a conjunction between the sources. Next we realized the algorithms and related codes that we implemented in the MATLAB environment. Our contribution lies in taking into account the imperfections (inaccuracies and uncertainties) linked to source information through the use of mass functions based on a simple Gaussian distribution support model in order to model each focal element independently of the others and to evaluate the belonging of a pixel to a class with respect to the majority of elements representing said class. The resulting results show that the DST approach is relatively satisfactory for the unsupervised classification of mineral surfaces and aquatic surfaces while it is not satisfactory for vegetated surfaces according to all proposed models. As for the DSmT, it presents satisfactory results for all the models proposed. The model with the exclusion integrity constraint E∩V ∩ M = φ was selected as the best model because having, in addition to an average rate of well-graded pixels of 93.34%, a compliance rate (96, 37%) with the terrain higher than those of the other models implemented.

Speaker verification using limited data is always a challenge for practical implementation as an application. An analysis on speaker verification studies for an i-vector based method using Mel-Frequency Cepstral Coefficient (MFCC) feature shows that the performance drops drastically as the duration of test data is reduced. This decrease in performance is due to insufficient phonetic coverage when we capture only the vocal tract feature. However the same can be improved if some source characteristics are taken into consideration. This paper attempts to improve the speaker verification performance using source characteristics. A recently proposed characterization of the voice source signal called the discrete cosine transform of the integrated linear prediction residual (DCTILPR) has been found to be useful as a speaker-specific feature. Speaker verification is performed over short test utterances in the NIST 2003 database using both the DCTILPR and MFCC features, and their score-level combination is found to give a significant performance improvement over the system using only the MFCC features.

The Quetta Syntaxis in western Baluchistan, Pakistan, is the result of an oroclinal bend of the western mountain belt and serves as a junction for different faults. As this area also lies close to the left-lateral strike-slip Chaman fault, which marks the boundary between the Indian and Eurasian plates, the resulting seismological behavior of this regime is very complex. In the region of the Quetta Syntaxis, close to the fold and thrust belt of the Sulaiman and Kirthar Ranges, an earthquake with a magnitude of 6.4 (Mw) occurred on October 28, 2008, which was followed by a doublet on the very next day. Six more shocks associated with these major events then occurred (one foreshock and five aftershocks), with moment magnitudes greater than 4. Numerous researchers have tried to explain the source of this sequence based on seismological, GPS, and Environmental Satellite (ENVISAT)/Advanced Synthetic Aperture Radar (ASAR) data. Here, we used Advanced Land Observing Satellite (ALOS)/Phased Array-type L-band Synthetic Aperture Radar (PALSAR) InSAR data sets from both ascending and descending orbits that allow us to more completely detect the deformation signals around the epicentral region. The results indicated that the shock sequence can be explained by two right-lateral and two left-lateral strike-slip faults that also included reverse slip. The right-lateral faults have a curved geometry. Moreover, whereas previous studies have explained the aftershock crustal deformation with a different fault source, we found that the same left-lateral segment of the conjugate fault was responsible for the aftershocks. We thus confirmed the complex surface deformation signals from the moderate-sized earthquake. Intra-plate crustal bending and shortening often seem to be accommodated as conjugate faulting, without any single preferred fault orientation. We also detected two possible landslide areas along with the crustal deformation pattern.

In this paper we report on a number of speaker identification experiments that assume a phonetic-oriented segmentation scheme exists such as to motivate the extraction of psychoacoustically-motivated phase and pitch related features. MFCC features are also considered for benchmarking. An emphasis is given to an innovative shift-invariant phase-related feature that is closely linked to the glottal source. A very simple statistical modeling is proposed and adapted in order to highlight the relative discrimination capabilities of different feature types. Results are presented for individual features and a discussion is also developed regarding possibilities of fusing features at the speaker modeling stage, or fusing distances at the speaker identification stage. Ferreira, A. First Experiments on Speaker Identification Combining a New Shift-invariant Phase-related Feature (NRD), MFCCs and F0 Information.

On 18 January 2017, the 2016–2017 central Italy seismic sequence reached the Campotosto area with four events with magnitude larger than 5 in three hours (major event MW 5.5). To study the slip behavior on the causative fault/faults we followed two different methodologies: (1) we use Interferometric Synthetic Aperture Radar (InSAR) interferograms (Sentinel-1 satellites) and Global Positioning System (GPS) coseismic displacements to constrain the fault geometry and the cumulative slip distribution; (2) we invert near-source strong-motion, high-sampling-rate GPS waveforms, and high-rate GPS-derived static offsets to retrieve the rupture history of the two largest events. The geodetic inversion shows that the earthquake sequence occurred along the southern segment of the SW-dipping Mts. Laga normal fault system with an average slip of about 40 cm and an estimated cumulative geodetic moment of 9.29 × 1017 Nm (equivalent to a MW~6). This latter estimate is larger than the cumulative seis...

BOLLETTINO SISMICO ITALIANO: ANALISYS OF EARLY AFTERSHOCKS OF THE 2016 MW 6.0 AMATRICE, MW 5.9 VISSO AND MW 6.5 NORCIA EARTHQUAKES IN CENTRAL ITALY B. Castello e Gruppo di Lavoro Bollettino Sismico Italiano (A. Nardi, A. Marchetti, F.M. Mele, A. Rossi P. Battelli, M. Berardi, C. Castellano, C. Melorio, G. Modica, M. Pirro, S. Spadoni, L. Arcoraci, G. Lozzi, A. Battelli, C. Thermes, M.G. Ciaccio, B. Castello, N. Pagliuca, A. Lisi, L. Pizzino, P. Baccheschi, A. Sciarra, A. Bono, C. Marcocci, V. Lauciani, A. Mandiello, S. Pintore, M. Quintiliani, A. Frepoli, D. Latorre, D. Cheloni, A.M. Lombardi, M. Vallocchia, B. Cantucci, L. Colini, S. Pinzi, L. Scognamiglio, A. Basili, F. Villani, G. D’Addezio, T. Sgroi, A. Smedile, C. Montuori, R. Tardini, R. Tozzi, S. Monna, M.T. Mariucci, R. Di Maro, L. Margheriti) Istituto Nazionale di Geofisica e Vulcanologia, INGV

Source and excitation modeling in FDTD formulation has a significant impact on the method performance and the required simulation time. Since the abrupt source introduction yields intensive numerical variations in whole computational domain, a generally accepted solution is to slowly introduce the source, using appropriate shaping functions in time. The main goal of the optimization presented in this paper is to find balance between two opposite demands: minimal required computation time and acceptable degradation of simulation performance. Reducing the time necessary for source activation and deactivation is an important issue, especially in design of microwave structures, when the simulation is intensively repeated in the process of device parameter optimization. Here proposed optimized source models are realized and tested within an own developed FDTD simulation environment.

In this paper, it is proposed to apply the Dempster-Shafer Theory (DST) or the theory of evidence to map vegetation, aquatic and mineral surfaces with a view to detecting potential areas of observation of outcrops of geological formations (rocks, breastplates, regolith, etc.). The proposed approach consists in aggregating information by using the DST. From pretreated Aster satellite images (geo-referencing, geometric correction and resampling at 15 m), newchannels were produced by determining the spectral indices NDVI, MNDWI and NDBaI. Then, the DST formalism was modeled and generated under the MATLAB software, an image segmented into six classes including three absolute classes (E,V,M) and three classes of confusion ({E,V}, {M,V}, {E,M}). The control on the land, based on geographic coordinates of pixels of different classes on said image, has made it possible to make a concordant interpretation thereof. Our contribution lies in taking into account imperfections (inaccuracies and uncertainties) related to source information by using mass functions based on a simple support model (two focal elements: the discernment framework and the potential set of belonging of the pixel to be classified) with a normal law for the good management of these.

This paper examines the association between the variability of the speech signal inside an analysis frame and the relative difficulty of classifying that frame. We introduce a novel measure of speech frame variability and show through classification experiments that this measure is a strong predictor of classifiability, even when conditioning on the distance to segment boundaries. Finally, we show how to incorporate the measure as weights in the discriminant function of a GMM-HMM recognizer, thereby increasing the relative importance of low variability frames in both decoding and training. This is shown to give a reduction in error rates.

I wish to thank my primary supervisor Prof. Michael Wagner for his introducing me to speech as a biometric, and for his support, suggestions and guidance throughout the learning process that has been my doctoral studies. Thank you also to Associate Prof. Roland Goecke for his active supervision, knowledge and advice over the last three years. My appreciation goes to Associate Prof.

In this paper, it is proposed to apply the Dempster-Shafer Theory (DST) or the theory of evidence to map vegetation, aquatic and mineral surfaces with a view to detecting potential areas of observation of outcrops of geological formations (rocks, breastplates, regolith, etc.). The proposed approach consists in aggregating information by using the DST. From pretreated Aster satellite images (geo-referencing, geometric correction and resampling at 15 m), newchannels were produced by determining the spectral indices NDVI, MNDWI and NDBaI. Then, the DST formalism was modeled and generated under the MATLAB software, an image segmented into six classes including three absolute classes (E,V,M) and three classes of confusion ({E,V}, {M,V}, {E,M}). The control on the land, based on geographic coordinates of pixels of different classes on said image, has made it possible to make a concordant interpretation thereof. Our contribution lies in taking into account imperfections (inaccuracies and uncertainties) related to source information by using mass functions based on a simple support model (two focal elements: the discernment framework and the potential set of belonging of the pixel to be classified) with a normal law for the good management of these.

Linked data continues to grow at a rapid rate, but a limitation of a lot of the data that is being published is the lack of a semantic description. There are tools, such as D2R, that allow a user to quickly convert a database into RDF, but these tools do not provide a way to easily map the data into an existing ontology. This paper presents a semiautomatic approach to map structured sources to ontologies in order to build semantic descriptions (source models). Since the precise mapping is sometimes ambiguous, we also provide a graphical user interface that allows a user to interactively refine the models. The resulting source models can then be used to convert data into RDF with respect to a given ontology or to define a SPARQL end point that can be queried with respect to an ontology. We evaluated the overall approach on a variety of sources and show that it can be used to quickly build source models with minimal user interaction.

Automatic voice pathology detection enables objective assessment of pathologies that affect the voice production mechanism. Detection systems have been developed using the traditional pipeline approach (consisting of the feature extraction part and the detection part) and using the modern deep learning-based end-to-end approach. Due to the lack of vast amounts of training data in the study area of pathological voice, the former approach is still a valid choice. In the existing detection systems based on the traditional pipeline approach, the mel-frequency cepstral coefficient (MFCC) features can be regarded as the defacto standard feature set. In this study, automatic voice pathology detection is investigated by comparing the performance of various MFCC variants derived by considering two factors: the input and the filterbank in the cepstrum computation. For the first factor, three inputs (the voice signal, the glottal source and the vocal tract) are compared. The glottal source and the vocal tract are estimated using the quasi-closed phase glottal inverse filtering method. For the second factor, the mel-frequency and linear-frequency filterbanks are compared. Experiments were conducted separately using six databases consisting of voices produced by speakers suffering from one of four disorders (dysphonia, Parkinson's disease, laryngitis, or heart failure) and by healthy speakers. Support vector machine (SVM) was used as the classifier. The results show that by combining mel-and linear-frequency cepstral coefficients derived from the glottal source and vocal tract, better overall detection accuracy was obtained compared to the defacto MFCC features derived from the voice signal. Furthermore, this combination provided comparable or better performance than four existing cepstral feature extraction techniques in clean and high signal-to-noise ratio (SNR) conditions. INDEX TERMS Voice disorders, glottal inverse filtering, support vector machine, cepstral coefficients. I. INTRODUCTION Voice pathologies arise either due to physical changes in the voice production mechanism (e.g., in the respiratory system, vocal folds, and vocal tract) [1], [2] or due to improper vocal use when the physical structure of the mechanism is normal (e.g., vocal fatigue or ventricular phonation) [3]-[5]. Examples of voice pathologies are dysarthria [7], dysphonia [8], vocal polyp [9], and developmental dysphasia [13]. Voice pathology may also indicate early neurodegenerative disease such as Parkinson's disease (PD) [10]-[12], [14]. Voice pathology detection refers to a technology to automatically The associate editor coordinating the review of this manuscript and approving it for publication was Shuihua Wang. distinguish normal voices from pathological voices by computer using the recorded voice signal. Existing voice pathology detection systems can be divided into two categories: traditional pipeline systems and modern end-to-end systems [15]. The traditional pipeline system consists of two components [15], [16]: the feature extraction part and the detection part. The feature extraction part tries to capture discriminative information from acoustic voice signal waveforms by representing this information in compressed forms using a set of pre-defined features. The feature sets reported in the literature for voice pathology detection can be grouped into four categories: (1) perturbation measures (such as jitter and shimmer); (2) spectral and cepstral measures

Speaker Recognition is a multi-disciplinary branch of biometrics that may be used for identification, verification, and classification of individual speakers, with the capability of tracking, detection, and segmentation by extension. Recently, a comprehensive book on all aspects of speaker recognition was published [1]. Therefore, here we are not concerned with details of the standard modeling which is and has been used for the recognition task. In contrast, we present a review of the most recent literature and briefly visit the latest techniques which are being deployed in the various branches of this technology.

In this paper, it is proposed to apply the Dempster-Shafer Theory (DST) or the theory of evidence to map vegetation, aquatic and mineral surfaces with a view to detecting potential areas of observation of outcrops of geological formations (rocks, breastplates, regolith, etc.). The proposed approach consists in aggregating information by using the DST. From pretreated Aster satellite images (geo-referencing, geometric correction and resampling at 15 m), newchannels were produced by determining the spectral indices NDVI, MNDWI and NDBaI. Then, the DST formalism was modeled and generated under the MATLAB software, an image segmented into six classes including three absolute classes (E,V,M) and three classes of confusion ({E,V}, {M,V}, {E,M}). The control on the land, based on geographic coordinates of pixels of different classes on said image, has made it possible to make a concordant interpretation thereof. Our contribution lies in taking into account imperfections (inaccuracies and uncertainties) related to source information by using mass functions based on a simple support model (two focal elements: the discernment framework and the potential set of belonging of the pixel to be classified) with a normal law for the good management of these.

In this paper, a novel excitation source-related feature set, viz., Teager Energy-based Mel Frequency Cepstral Coefficients (T-MFCC) is proposed for the task of spoken keyword detection. Experiments are carried out on TIMIT database for spoken keyword detection. Furthermore, state-of-the-art feature set, viz., MFCC is used as the baseline spectral feature set to represent implicitly vocal tract (i.e., system) information. The idea is to exploit the vocal-source (and its nonlinear coupling with formant) and system-related information embedded in the spoken query. Experimental results show % EER of 17.23 and 22.58 for MFCC and proposed T-MFCC features, respectively. However, the significant reduction in % EER, i.e., by 1.8 % (as compared to MFCC) is observed when evidences from T-MFCC and MFCC are combined using score-level fusion; indicating that proposed feature set captures complementary linguistic information (in the spoken keyword) than MFCC alone.

In this paper, we elaborate on mobile phone identification from recorded speech signals. The goal is to extract intrinsic traces related to the mobile phone used to record a speech signal. Mel frequency cepstral coefficients (MFCCs) are extracted from any recorded speech signal at a frame level. The sequences of the MFCC vectors extracted from each recording device train a Gaussian Mixture Model with diagonal covariance matrices. A Gaussian supervector is derived by concatenating the mean vectors and the main diagonals of the covariance matrices that is used as a template for each device. Experiments were conducted on a database of 21 mobile phones of various models from 7 different brands. The aforementioned database, that is called MOBIPHONE, was collected by recording 10 utterances, uttered by 12 male speakers and another 12 female speakers, randomly chosen from the TIMIT database. Three commonly used classifiers were employed, such as Support Vector Machines with different kernels, a Radial Basis Functions neural network, and a Multi-Layer Perceptron. The best identification accuracy (97.6%) was obtained by the Radial Basis Functions neural network.

Motivated by the speaker-specificity and stationarity of subglottal acoustics, this paper investigates the utility of subglottal cepstral coefficients (SGCCs) for speaker identification (SID) and verification (SV). SGCCs can be computed using accelerometer recordings of subglottal acoustics, but such an approach is infeasible in real-world scenarios. To estimate SGCCs from speech signals, we adopt the Bayesian minimum mean squared error (MMSE) estimator proposed in the speech-to-articulatory inversion literature. The joint distribution of SGCCs and speech MFCCs is modeled using the WashU-UCLA corpus (containing simultaneous recordings of speech and subglottal acoustics), and the resulting model is used to obtain an MMSE estimate of SGCCs from unseen (test) MFCCs. Cross-validation experiments on the WashU-UCLA corpus show that the estimation efficacy, on average, is speaker dependent. A score-level fusion of MFCC and SGCC systems outperforms the MFCC-only baseline in both SID and SV tasks. On the TIMIT database (SID), the relative reduction in identification error is 16, 40 and 51% for G.712-filtered (300-3400 Hz), narrowband (0-4000 Hz) and wideband (0-8000 Hz) speech, respectively. On the NIST 2008 database (SV), the relative reduction in equal error rate is 4 and 11% for 10 and 5 second utterances, respectively.

In this paper, it is proposed to apply the Dempster-Shafer Theory (DST) or the theory of evidence to map vegetation, aquatic and mineral surfaces with a view to detecting potential areas of observation of outcrops of geological formations (rocks, breastplates, regolith, etc.). The proposed approach consists in aggregating information by using the DST. From pretreated Aster satellite images (geo-referencing, geometric correction and resampling at 15 m), newchannels were produced by determining the spectral indices NDVI, MNDWI and NDBaI. Then, the DST formalism was modeled and generated under the MATLAB software, an image segmented into six classes including three absolute classes (E,V,M) and three classes of confusion ({E,V}, {M,V}, {E,M}). The control on the land, based on geographic coordinates of pixels of different classes on said image, has made it possible to make a concordant interpretation thereof. Our contribution lies in taking into account imperfections (inaccuracies and uncertainties) related to source information by using mass functions based on a simple support model (two focal elements: the discernment framework and the potential set of belonging of the pixel to be classified) with a normal law for the good management of these.

Semantic models of data sources and services provide support to automate many tasks such as source discovery, data integration, and service composition, but writing these semantic descriptions by hand is a tedious and time-consuming task. Most of the related work focuses on automatic annotation with classes or properties of source attributes or input and output parameters. However, constructing a source model that includes the relationships between the attributes in addition to their semantic types remains a largely unsolved problem. In this paper, we present a graph-based approach to hypothesize a rich semantic description of a new target source from a set of known sources that have been modeled over the same domain ontology. We exploit the domain ontology and the known source models to build a graph that represents the space of plausible source descriptions. Then, we compute the top k candidates and suggest to the user a ranked list of the semantic models for the new source. The approach takes into account user corrections to learn more accurate semantic descriptions of future data sources. Our evaluation shows that our method produces models that are twice as accurate than the models produced using a state of the art system that does not learn from prior models.

On 18 January 2017, the 2016–2017 central Italy seismic sequence reached the Campotosto area with four events with magnitude larger than 5 in three hours (major event MW 5.5). To study the slip behavior on the causative fault/faults we followed two different methodologies: (1) we use Interferometric Synthetic Aperture Radar (InSAR) interferograms (Sentinel-1 satellites) and Global Positioning System (GPS) coseismic displacements to constrain the fault geometry and the cumulative slip distribution; (2) we invert near-source strong-motion, high-sampling-rate GPS waveforms, and high-rate GPS-derived static offsets to retrieve the rupture history of the two largest events. The geodetic inversion shows that the earthquake sequence occurred along the southern segment of the SW-dipping Mts. Laga normal fault system with an average slip of about 40 cm and an estimated cumulative geodetic moment of 9.29 × 1017 Nm (equivalent to a MW~6). This latter estimate is larger than the cumulative seis...

We propose a practical, feature-level fusion approach for combining acoustic and articulatory information in speaker verification task. We find that concatenating articulation features obtained from the measured speech production data with conventional Mel-frequency cepstral coefficients (MFCCs) improves the overall speaker verification performance. However, since access to the measured articulatory data is impractical for real world speaker verification applications, we also experiment with estimated articulatory features obtained using acoustic-to-articulatory inversion technique. Specifically, we show that augmenting MFCCs with articulatory features obtained from subject-independent acoustic-to-articulatory inversion technique also significantly enhances the speaker verification performance. This performance boost could be due to the information about inter-speaker variation present in the estimated articulatory features, especially at the mean and variance level. Experimental results on the Wisconsin X-Ray Microbeam database show that the proposed acoustic-estimatedarticulatory fusion approach significantly outperforms the traditional acoustic-only baseline, providing up to 10% relative reduction in Equal Error Rate (EER). We further show that we can achieve an additional 5% relative reduction in EER after score-level fusion.

Speaker identification is a well-established research problem but has not been a major application used in gaming scenarios. In this paper, we propose a new algorithm for the open-set, text-independent, speaker ID problem, applied as an important component (among other cues) of a game player identification system. This scenario poses new challenges: far-field, limited training and very short test data, and almost real-time processing. To tackle this, we introduce new and more informative feature sets. The scores given by these feature sets are then combined in an optimal way to construct the final score. Experimental results on the gaming device's processed reverberated-speech show the effectiveness of the new features, and that reliable decisions can be made after very short (2-5 second) test utterances required by the gaming scheme.

In this paper, a novel excitation source-related feature set, viz., Teager Energy-based Mel Frequency Cepstral Coefficients (T-MFCC) is proposed for the task of spoken keyword detection. Experiments are carried out on TIMIT database for spoken keyword detection. Furthermore, state-of-the-art feature set, viz., MFCC is used as the baseline spectral feature set to represent implicitly vocal tract (i.e., system) information. The idea is to exploit the vocal-source (and its nonlinear coupling with formant) and system-related information embedded in the spoken query. Experimental results show % EER of 17.23 and 22.58 for MFCC and proposed T-MFCC features, respectively. However, the significant reduction in % EER, i.e., by 1.8 % (as compared to MFCC) is observed when evidences from T-MFCC and MFCC are combined using score-level fusion; indicating that proposed feature set captures complementary linguistic information (in the spoken keyword) than MFCC alone.

In this paper, it is proposed to apply the Dempster-Shafer Theory (DST) or the theory of evidence to map vegetation, aquatic and mineral surfaces with a view to detecting potential areas of observation of outcrops of geological formations (rocks, breastplates, regolith, etc.). The proposed approach consists in aggregating information by using the DST. From pretreated Aster satellite images (geo-referencing, geometric correction and resampling at 15 m), newchannels were produced by determining the spectral indices NDVI, MNDWI and NDBaI. Then, the DST formalism was modeled and generated under the MATLAB software, an image segmented into six classes including three absolute classes (E,V,M) and three classes of confusion ({E,V}, {M,V}, {E,M}). The control on the land, based on geographic coordinates of pixels of different classes on said image, has made it possible to make a concordant interpretation thereof. Our contribution lies in taking into account imperfections (inaccuracies and uncertainties) related to source information by using mass functions based on a simple support model (two focal elements: the discernment framework and the potential set of belonging of the pixel to be classified) with a normal law for the good management of these.

In this paper, it is proposed to apply the Dempster-Shafer Theory (DST) or the theory of evidence to map vegetation, aquatic and mineral surfaces with a view to detecting potential areas of observation of outcrops of geological formations (rocks, breastplates, regolith, etc.). The proposed approach consists in aggregating information by using the DST. From pretreated Aster satellite images (geo-referencing, geometric correction and resampling at 15 m), newchannels were produced by determining the spectral indices NDVI, MNDWI and NDBaI. Then, the DST formalism was modeled and generated under the MATLAB software, an image segmented into six classes including three absolute classes (E,V,M) and three classes of confusion ({E,V}, {M,V}, {E,M}). The control on the land, based on geographic coordinates of pixels of different classes on said image, has made it possible to make a concordant interpretation thereof. Our contribution lies in taking into account imperfections (inaccuracies and uncertainties) related to source information by using mass functions based on a simple support model (two focal elements: the discernment framework and the potential set of belonging of the pixel to be classified) with a normal law for the good management of these.

Efficient computation of scalar optical diffraction field due to an object is an essential issue in holographic 3D television systems. The first step in the computation process is to construct an object. As a solution for this step, we assume that an object can be represented by a set of distributed data points over a space. The second step is to determine which algorithm provides better performance. The source model whose performance is investigated is based on superposition of the diffraction fields emanated from the hypothetical light sources located at the given sample points. Its performance is evaluated according to visual quality of the reconstructed field and its algorithmic complexity. Source model provides acceptable reconstructed patterns when the region in which the samples are given has a narrow depth along the longitudinal direction and a wide extent along the transversal directions. Also, the source model gives good results when the cumulative field at the location of each point due to all other sources tends to be independent of that location.

This paper describes an approach of speech recognition by using the Mel-Scale Frequency Cepstral Coefficients (MFCC) extracted from speech signal of spoken words. Principal Component Analysis is employed as the supplement in feature dimensional reduction state, prior to training and testing speech samples via Maximum Likelihood Classifier (ML) and Support Vector Machine (SVM). Based on experimental database of total 40 times of speaking words collected under acoustically controlled room, the sixteen-ordered MFCC extracts have shown the improvement in recognition rates significantly when training the SVM with more MFCC samples by randomly selected from database, compared with the ML.

Usually, speaker recognition systems do not take into account the short-term dependence between the vocal source and the vocal tract. A feasibility study that retains this dependence is presented here. A model of joint probability functions of the pitch and the feature vectors is proposed. Three strategies are designed and compared for all female speakers taken from the SPIDRE corpus. The first operates on all voiced and unvoiced speech segments (baseline strategy). The second strategy considers only the voiced speech segments and the last includes the short-term pitch information along with the standard MFCC. We use two pattern recognizers: LVQ-SLP and GMM. In all cases, we observe an increase in the identification rates and more specifically when using a time duration of 500 ms (6% higher).

Automatic speaker recognition system is used to recognize an unknown speaker among several reference speakers by making use of speaker-specific information from their speech. In this paper, we introduce a novel, hierarchical, text-independent speaker recognition. Our baseline speaker recognition system accuracy, built using statistical modeling techniques, gives an accuracy of 81 % on the standard MIT database. We then propose and implement a novel state-space pruning technique by performing gender recognition before speaker recognition so as to improve the accuracy/timeliness of our baseline speaker recognition system. Based on the experiments conducted on the MIT database, we demonstrate that our proposed system improves the accuracy over the baseline system by approximately 2%, while reducing the computational time by more than 30%.

Our initial speaker verification study exploring the impact of mismatch in training and test conditions finds that the mismatch in sensor and acoustic environment results in significant performance degradation compared to other mismatches like language and style (Haris et al. in Int. J. Speech Technol., 2012). In this work we present a method to suppress the mismatch between the training and test speech, specifically due to sensor and acoustic environment. The method is based on identifying and emphasizing more speaker specific and less mismatch affected vowellike regions (VLRs) compared to the other speech regions. VLRs are separated from the speech regions (regions detected using voice activity detection (VAD)) using VLR onset point (VLROP) and are processed independently during training and testing of the speaker verification system. Finally, the scores are combined with more weight to that generated by VLRs as those are relatively more speaker specific and less mismatch affected. Speaker verification studies are conducted using the mel-frequency cepstral coefficients (MFCCs) as feature vectors. The speaker modeling is done using the Gaussian mixture model-universal background model and the state-of-the-art i-vector based approach. The experimental results show that for both the systems, proposed approach provides consistent performance improve

Abstract—The crosslingual voice conversion problem refers to the replacement of a speaker’s timbre or vocal identity in a recorded sentence, assuming that the source speaker and target speaker use different languages. This problem differs from typical voice conversion in the sense that the mapping of acoustical features cannot depend on time-aligned recordings of source and target speakers uttering the same sentences. This paper presents an overview of a general crosslingual voice conversion system and discusses the most important techniques used in each step of the conversion process. Keywords-Crosslingual Voice Conversion; Timbre transformation; Vocal Identity;

This study proposes a new method of fitting a glottal model to the glottal flow estimate using system identification (SI) algorithms. Each period of the glottal estimate is split into open and closed phases and each phase is modelled as the output of a linear filter. This approach allows the parametric model fitting task to be cast as a system identification problem and sidesteps issues encountered with standard glottal parametrisation algorithms. The study compares the performance of two SI methods: Steiglitz-McBride and Prony. The tests were performed on synthetic glottal signals (n=121) and real speech (n=50 healthy, n=23 pathological). The effectiveness of the techniques is quantified by calculating the Normalised Root Mean Squared Error (NRMSE) between the estimated glottal fit and the glottal estimate. Tests on synthetic glottal signals show that the average performance of the Steiglitz-McBride method (97.25%) was better than the Prony method (70.41%). Real speech tests produced results of 64.29% and 51.57% for healthy and pathological speech respectively. The results show that system identification techniques can produce robust parametric model estimates of the glottal waveform and that the Steiglitz-McBride method is superior to the Prony method for this task.

The possibility to discriminate between speech and music signals by using a feature based on low frequency modulation has been investigated. Three different low frequency modulation parameters have been extracted and tested concerning the ability of discrimination. The low frequency modulation amplitudes calculated over 20 critical bands and their standard deviations were found to be good features for this discrimination task even with VQ models. They were also found to be less sensitive to channel quality and model size than MFCC features. 1. Background and review In preprocessors for speech recognition there is often a need to classify and segment the signal before the transcription. Many attempts have been made to discriminate between speech and music or other sound classes [1-4]. Most of these are based on knowledge of the speech production and perception or a combination of the two. The relation between these is described by Greenberg [5] where he in particular points out the i...

Various algorithms for text-independent speaker recognition have been developed through the decades, aiming to improve both accuracy and efficiency. This paper presents a novel PCA/LDA-based approach that is faster than traditional statistical model-based methods and achieves competitive results. First, the performance based on only PCA and only LDA is measured; then a mixed model, taking advantages of both methods, is introduced. A subset of the TIMIT corpus composed of 200 male speakers, is used for enrollment, validation and testing. The best results achieve 100%, 96% and 95% classification rate at population level 50, 100 and 200, using 39- dimensional MFCC features with delta and double delta. These results are based on 12-second text-independent speech for training and 4-second data for test. These are comparable to the conventional MFCC-GMM methods, but require significantly less time to train and operate.

The cross lingual voice conversion problem refers to the replacement of a speaker's timbre or vocal identity in a recorded sentence, assuming that the source speaker and target speaker use different languages. This problem differs from typical voice conversion in the sense that the mapping of acoustical features cannot depend on time-aligned recordings of source and target speakers uttering the

Various algorithms for text-independent speaker recognition have been developed through the decades, aiming to improve both accuracy and efficiency. This paper presents a novel PCA/LDA-based approach that is faster than traditional statistical model-based methods and achieves competitive results. First, the performance based on only PCA and only LDA is measured; then a mixed model, taking advantages of both methods, is introduced. A subset of the TIMIT corpus composed of 200 male speakers, is used for enrollment, validation and testing. The best results achieve 100%, 96% and 95% classification rate at population level 50, 100 and 200, using 39dimensional MFCC features with delta and double delta. These results are based on 12-second text-independent speech for training and 4-second data for test. These are comparable to the conventional MFCC-GMM methods, but require significantly less time to train and operate.

VAD is a reason for the trouble of discrimination between external noise and voice. VAD is an issue and for that reason various techniques have been suggested. Some are based upon power spectral density derived characteristics, and others utilize the signal periodicity. The goal of this paper is analyzing the combined utilization of resource and also filter functionality. A shared information based analysis demonstrates the source related functions, especially the proposed people, are much more discriminatory. The qualities are in addition of the input of a multi condition version education, synthetic neural community classifier. The integration of resource and also filter info is suggested as two approaches: the fusion of decision and function. Our tests demonstrate that the corresponding mistake speed of choice fusion is definitely decreased by 3%. In relation to the 150 minutes of information collected in world that is real, the last program proposed is when compared with 4 state of the art strategies. The program proposed results in a big improvement in reliability under all the conditions probably in the most complex VAD (twenty four per cent total, average) because of its strong source features, its multi condition instruction and also its highly effective info fusion.

We study robust pitch synchronous parameters that are derived from envelope and instantaneous frequencies estimated via a bank of cochlear filters. Closed set Speaker Identification experiments are performed on the SPIDRE corpus with matched and mis- matched handsets conditions. The recognizer is based on a hybrid Linear Vector Quantization and Single Layer Perceptron (LVQ- SLP). Experiments are reported with different

In the source-filter model of speech production, physiologically, the source corresponds to the vocal fold vibrations and the filter corresponds to the spectrum-shaping vocal tract. Vocal tract-based features like the mel-frequency cepstral coefficients (MFCCs) have been shown to contain speaker information. However, voice source (VS)-based features have also been shown to perform well in speaker recognition tasks, thereby revealing that the VS does contain speaker information. Moreover, a combination of the vocal tract and VS-based features has been shown to give an improved performance, showing that the latter contains supplementary speaker information.
In this study, the existing techniques for extracting speaker information from the VS are reviewed, and it is observed that parametric features perform poorly than non-parametric features. Here, an attempt is made to propose an alternate way of characterizing the VS to extract speaker information, and to study the merits and shortcomings of the proposed speaker-specific features.
The integrated linear prediction residual (ILPR) is used as the VS estimate. It is hypothesized here that a speaker’s voice may be characterized by the relative proportions of the harmonics present in the VS. The pitch synchronous discrete cosine transform (DCT) is shown to capture these, and the gross shape of the ILPR in a few coefficients. The ILPR and hence its DCT coefficients are visually observed to have both inter and intra-speaker variability, and thus it is hypothesized that the distribution of the DCT coefficients may capture speaker information, and this distribution is modeled by a Gaussian mixture model (GMM).
The DCT coefficients of the ILPR (termed the DCTILPR) are directly used as a feature vector in speaker identification (SID) tasks. By conducting SID experiments on three standard databases, it is found that the proposed DCTILPR features fare comparably with the existing VS-based features. It is also found that the gross shape of the VS contains most of the speaker information, and the very fine structure of the VS does not help in distinguishing speakers, and instead leads to more confusion between speakers. The major drawbacks of the DCTILPR are the session and handset variability, but they are also present in the existing state-of-the-art speaker-specific VS-based features and the MFCCs, and hence seem to be common problems. There are techniques to compensate these variabilities, which need to be used when the systems using these features are deployed in an actual application.
The DCTILPR is found to improve the SID accuracy of a system trained with MFCC features by 12%, indicating that the DCTILPR features capture speaker information which is missed by the MFCCs. It is also found that a combination of MFCC and DCTILPR features on a speaker verification task gives significant performance improvement in the case of short test utterances. Thus, on the whole, this study proposes an alternate way of extracting speaker information from the VS, and adds to the evidence for speaker information present in the VS.

ABSTRACT This paper addresses the problem of pitch modification, as an important module for an efficient voice transformation system. The Deterministic plus Stochastic Model of the residual signal we proposed in a previous work is compared to TDPSOLA, HNM and STRAIGHT. The four methods are compared through an important subjective test. The influence of the speaker gender and of the pitch modification ratio is analyzed.

Abstract Voice source analysis and modelling has played a key role in important speech applications such as speech recognition, speech synthesis and speaker recognition. This work presents a robust algorithm for glottal closure detection and a novel set of voice source features for speaker recognition. In the first part of the dissertation the DYPSA algorithm is developed for detecting glottal closure instants (GCIs).

Abstract This paper presents a data-driven approach to the modelling of voice source waveforms. The voice source is a signal that is estimated by inverse-filtering speech signals with an estimate of the vocal tract filter. It is used in speech analysis, synthesis, recognition and coding to decompose a speech signal into its source and vocal tract filter components. Existing approaches parameterize the voice source signal with physically-or mathematically-motivated models.

Recent years have seen an explosion of interest in using neural oscillations to characterize the mechanisms supporting cognition and emotion. Oftentimes, oscillatory activity is indexed by mean power density in predefined frequency bands. Some investigators use broad bands originally defined by prominent surface features of the spectrum. Others rely on narrower bands originally defined by spectral factor analysis (SFA). Presently, the robustness and sensitivity of these competing band definitions remains unclear. Here, a Monte Carlo-based SFA strategy was used to decompose the tonic (“resting” or “spontaneous”) electroencephalogram (EEG) into five bands: delta (1–5 Hz), alpha-low (6–9 Hz), alpha-high (10–11 Hz), beta (12–19 Hz), and gamma (N21 Hz). This pattern was consistent across SFA methods, artifact correction/rejection procedures, scalp regions, and samples. Subsequent analyses revealed that SFA failed to deliver enhanced sensitivity; narrow alpha sub-bands proved no more sensitive than the classical broadband to individual differences in temperament or mean differences in task-induced activation. Other analyses suggested that residual ocular and muscular artifact was the dominant source of activity during quiescence in the delta and gamma bands. This was observed following threshold-based artifact rejection or independent component analysis (ICA)- based artifact correction, indicating that such procedures do not necessarily confer adequate protection. Collectively, these findings highlight the limitations of several commonly used EEG procedures and underscore the necessity of routinely performing exploratory data analyses, particularly data visualization, prior to hypothesis testing. They also suggest the potential benefits of using techniques other than SFA for interrogating high-dimensional EEG datasets in the frequency or time–frequency (event-related spectral perturbation, event-related synchronization/desynchronization) domains. KEY WORDS: principal components analysis (PCA); exploratory factor analysis (EFA); blind source separation (BSS); resting neural activity; resting EEG; frontal alpha asymmetry; frontal EEG asymmetry.