In this paper, we explore the possibility of using WaveNet architecture as a statistical vocoder. that case, generation speech waveforms is locally conditioned only by acoustic features. Focusing on single speaker case at moment, investigate impact local conditions well amount data available for training. Furthermore, variations are considered and discussed in context our work. We compare work against very recent which also used vocoder same data. More specifically, two female male speakers...

Recently, generative neural network models which operate directly on raw audio, such as WaveNet, have improved the state of art in text-to-speech synthesis (TTS). Moreover, there is increasing interest using these statistical vocoders for generating speech waveforms from various acoustic features. However, also a need to reduce model complexity, without compromising quality. Previously, glottal pulseforms (i.e., time-domain corresponding source human voice production mechanism) been...

<para xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> Epilepsy is one of the most common brain disorders and may result in dysfunction cognitive disturbances. Epileptic seizures usually begin childhood without being accommodated by damage are tolerated drugs that produce no dysfunction. In this study, function evaluated children with mild epileptic controlled antiepileptic drugs. Under prism, we propose a concise technical framework combining...

Recent speech technology research has seen a growing interest in using WaveNets as statistical vocoders, i.e., generating waveforms from acoustic features.These models have been shown to improve the generated quality over classical vocoders many tasks, such text-to-speech synthesis and voice conversion.Furthermore, conditioning with features allows sharing waveform generator model across multiple speakers without additional speaker codes.However, multi-speaker WaveNet require large amounts...

Although Electroencephalographic (EEG) signal synchronization studies have been a topic of increasing interest lately, there is no similar effort in the visualization such measures. In this direction graph-theoretic approach devised to study and stress coupling dynamics task-performing dynamical networks proposed. Both linear nonlinear interdependence measures are investigated an alcoholism paradigm during mental rehearsal pictures, which known reflect impairment. More specifically, widely...

Over the past few years there has been an increased interest in studying underlying neural mechanism of cognitive brain activity. In this direction, we study activity based on its independent components instead EEG signal itself. Both linear and nonlinear synchronization measures are applied to components, which free volume conduction effects background noise. More specifically, a robust state-space generalized assessment method recently introduced partial directed coherence investigated...

In this paper, we suggest a new parallel, non-causal and shallow waveform domain architecture for speech enhancement based on FFTNet, neural network generating high quality audio waveform. contrast to other approaches like WaveNet, FFTNet uses an initial wide dilation pattern. Such better represents the long term correlated structure of in time domain, where noise is usually highly non-correlated, therefore it suitable enhancement. To further strengthen feature architecture, present sample...

The aims were to develop and validate a VO 2peak prediction equation from treadmill running test in active male adolescents. Eighty-eight athletes (12–18 yrs.) performed maximal exercise on assess the actual 20m Shuttle-Run-Test (20mST). A step-wise linear regression analysis was used following for estimation of (mL·kg −1 ·min ) = 35.477 + 1.832 × duration min - 0.010 body mass kg developed. cross-validation statistics were: R .54, CE 0.1 mL·kg , SEE 2.5 (4.6%), TE 2.6 (4.9%). values (CE,...

In this paper we study the detection of hesitation filled pauses in oral presentations university lectures taught Greek language and recorded using a tablet PC via specialized software. We suggest hierarchical approach fusing video data with audio for increasing precision rate our system. The method works at frame level rather than usual segmental more accurate synchronization after removing detected hesitations. Audio characteristics are modeled Gaussian Mixture Models while stationarity is...

Gene Ontology information related to the biological role of genes is organized in a hierarchical manner that can be represented by directed acyclic graph (DAG). Space filling visualizations, such as treemaps, have capacity display thousands items legibly limited space via two-dimensional rectangular map. Treemaps been used visualize first transforming DAG into tree. However this transformation has several undesirable effects producing trees with large number nodes and scattering rectangles...

This paper presents BrainNetVis, a tool which serves brain network modelling and visualization, by providing both quantitative qualitative measures of interconnectivity. It emphasizes the needs that led to creation this presenting similar works in field describing how our contributes existing scenery. also describes methods used for calculation graph metrics (global vertex metrics), carry information. To make clear understandable, we use an exemplar dataset throughout paper, on calculations...

BrainNetVis is an application, written in Java, that displays and analyzes synchronization networks from brain signals. The program implements a number of network indices visualization techniques. We demonstrate its use through case study left hand foot motor imagery. data sets were provided by the Berlin BCI group. Using this we managed to find differences between average comparing them with idle state network.

Hidden Markov models (HMMs) are becoming the dominant approach for text-to-speech synthesis (TTS). HMMs provide an attractive acoustic modeling scheme which has been exhaustively investigated and developed many years. Modern HMM-based speech synthesizers have approached quality of best state-of-the-art unit selection systems. However, we believe that statistical parametric not reached its potential, since limited by several assumptions do apply to properties speech. We, therefore, propose in...

Linear Dynamical Models (LDMs) have been used in speech synthesis recently as an alternative to hidden Markov models (HMMs). Among the advantages of LDMs are ability capture dynamics and achievement synthesized quality similar HMM-based systems on a smaller footprint. However, such HMM case, produce over-smoothed trajectories parameters, resulting muffled synthetic speech. Inspired by problem found synthesis, where naturalness is greatly improved when global variance (GV) compensated, this...

Low speech intelligibility in noisy listening conditions makes more difficult our communication with others.Various strategies have been suggested to modify a signal before it is presented environment the goal increase its intelligibility.A state-of-the art approach, referred as Spectral Shaping and Dynamic Range Compression (SS-DRC), relies on modifying spectral temporal structure of clean has shown considerably improve conditions.In this paper, we present non-causal Wavenet-like model for...

DAGmaps are space lling visualizations of DAGs that generalize treemaps. Deciding whether or not a DAG admits DAGmap is NPcomplete. Although any layered planar one-dimensional there was no complete characterization the class admit DAGmap. In this paper we prove if and only it directed -visibility representation. Then characterize representations. This consists downward straight-line drawing such all source sink vertices assigned to external face. Finally show denes three-dimensional