scholarly journals An Analysis of the Short Utterance Problem for Speaker Characterization

2019 ◽  
Vol 9 (18) ◽  
pp. 3697 ◽  
Author(s):  
Ignacio Viñals ◽  
Alfonso Ortega ◽  
Antonio Miguel ◽  
Eduardo Lleida
Keyword(s):  
Deep Neural Network ◽  
State Of The Art ◽  
Significant Loss ◽  
Point Of View ◽  
Speaker Variability ◽  
Phonetic Similarity ◽  
Short Time ◽  
Short Utterances ◽  
Time Information ◽  
Short Utterance

Speaker characterization has always been conditioned by the length of the evaluated utterances. Despite performing well with large amounts of audio, significant degradations in performance are obtained when short utterances are considered. In this work we present an analysis of the short utterance problem providing an alternative point of view. From our perspective the performance in the evaluation of short utterances is highly influenced by the phonetic similarity between enrollment and test utterances. Both enrollment and test should contain similar phonemes to properly discriminate, being degraded otherwise. In this study we also interpret short utterances as incomplete long utterances where some acoustic units are either unbalanced or just missing. These missing units are responsible for the speaker representations to be unreliable. These unreliable representations are biased with respect to the reference counterparts, obtained from long utterances. These undesired shifts increase the intra-speaker variability, causing a significant loss of performance. According to our experiments, short utterances (3–60 s) can perform as accurate as if long utterances were involved by just reassuring the phonetic distributions. This analysis is determined by the current embedding extraction approach, based on the accumulation of local short-time information. Thus it is applicable to most of the state-of-the-art embeddings, including traditional i-vectors and Deep Neural Network (DNN) xvectors.

scholarly journals Neuro-Inspired Computing with Spin-VCSELs

2021 ◽  
Vol 11 (9) ◽  
pp. 4232
Author(s):  
Krishan Harkhoe ◽  
Guy Verschaffelt ◽  
Guy Van der Sande
Keyword(s):  
Processing Speed ◽  
Delay Time ◽  
High Speed ◽  
State Of The Art ◽  
Polarization Modulation ◽  
Time Interval ◽  
Computing Technique ◽  
High Modulation ◽  
Short Time ◽  
Modulation Speed

Delay-based reservoir computing (RC), a neuromorphic computing technique, has gathered lots of interest, as it promises compact and high-speed RC implementations. To further boost the computing speeds, we introduce and study an RC setup based on spin-VCSELs, thereby exploiting the high polarization modulation speed inherent to these lasers. Based on numerical simulations, we benchmarked this setup against state-of-the-art delay-based RC systems and its parameter space was analyzed for optimal performance. The high modulation speed enabled us to have more virtual nodes in a shorter time interval. However, we found that at these short time scales, the delay time and feedback rate heavily influence the nonlinear dynamics. Therefore, and contrary to other laser-based RC systems, the delay time has to be optimized in order to obtain good RC performances. We achieved state-of-the-art performances on a benchmark timeseries prediction task. This spin-VCSEL-based RC system shows a ten-fold improvement in processing speed, which can further be enhanced in a straightforward way by increasing the birefringence of the VCSEL chip.

scholarly journals Fast Accurate and Automatic Brushstroke Extraction

2021 ◽  
Vol 17 (2) ◽  
pp. 1-24
Author(s):  
Yunfei Fu ◽  
Hongchuan Yu ◽  
Chih-Kuo Yeh ◽  
Tong-Yee Lee ◽  
Jian J. Zhang
Keyword(s):  
Neural Network ◽  
Efficient Algorithm ◽  
Deep Neural Network ◽  
High Efficiency ◽  
State Of The Art ◽  
High Reliability ◽  
The Other ◽  
Manual Annotation ◽  
Stroke Extraction ◽  
Art Research

Brushstrokes are viewed as the artist’s “handwriting” in a painting. In many applications such as style learning and transfer, mimicking painting, and painting authentication, it is highly desired to quantitatively and accurately identify brushstroke characteristics from old masters’ pieces using computer programs. However, due to the nature of hundreds or thousands of intermingling brushstrokes in the painting, it still remains challenging. This article proposes an efficient algorithm for brush Stroke extraction based on a Deep neural network, i.e., DStroke. Compared to the state-of-the-art research, the main merit of the proposed DStroke is to automatically and rapidly extract brushstrokes from a painting without manual annotation, while accurately approximating the real brushstrokes with high reliability. Herein, recovering the faithful soft transitions between brushstrokes is often ignored by the other methods. In fact, the details of brushstrokes in a master piece of painting (e.g., shapes, colors, texture, overlaps) are highly desired by artists since they hold promise to enhance and extend the artists’ powers, just like microscopes extend biologists’ powers. To demonstrate the high efficiency of the proposed DStroke, we perform it on a set of real scans of paintings and a set of synthetic paintings, respectively. Experiments show that the proposed DStroke is noticeably faster and more accurate at identifying and extracting brushstrokes, outperforming the other methods.

scholarly journals Sustainable Agri-Food Processes and Circular Economy Pathways in a Life Cycle Perspective: State of the Art of Applicative Research

2021 ◽  
Vol 13 (5) ◽  
pp. 2472
Author(s):  
Teodora Stillitano ◽  
Emanuele Spada ◽  
Nathalie Iofrida ◽  
Giacomo Falcone ◽  
Anna Irene De Luca
Keyword(s):  
Life Cycle Assessment ◽  
Life Cycle ◽  
Circular Economy ◽  
Social Life ◽  
Food Systems ◽  
State Of The Art ◽  
Terrestrial Ecosystems ◽  
Point Of View ◽  
Social Life Cycle Assessment ◽  
Qualitative Synthesis

This study aims at providing a systematic and critical review on the state of the art of life cycle applications from the circular economy point of view. In particular, the main objective is to understand how researchers adopt life cycle approaches for the measurement of the empirical circular pathways of agri-food systems along with the overall lifespan. To perform the literature review, the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) protocol was considered to conduct a review by qualitative synthesis. Specifically, an evaluation matrix has been set up to gather and synthesize research evidence, by classifying papers according to several integrated criteria. The literature search was carried out employing scientific databases. The findings highlight that 52 case studies out of 84 (62% of the total) use stand-alone life cycle assessment (LCA) to evaluate the benefits/impacts of circular economy (CE) strategies. In contrast, only eight studies (9.5%) deal with the life cycle costing (LCC) approach combined with other analyses while no paper deals with the social life cycle assessment (S-LCA) methodology. Global warming potential, eutrophication (for marine, freshwater, and terrestrial ecosystems), human toxicity, and ecotoxicity results are the most common LCA indicators applied. Only a few articles deal with the CE assessment through specific indicators. We argue that experts in life cycle methodologies must strive to adopt some key elements to ensure that the results obtained fit perfectly with the measurements of circularity and that these can even be largely based on a common basis.

New Trends in Asymptotic Approaches: Summation and Interpolation Methods

2001 ◽  
Vol 54 (1) ◽  
pp. 69-92 ◽  
Author(s):  
Igor V. Andrianov ◽  
Jan Awrejcewicz
Keyword(s):  
Applied Sciences ◽  
Solid Mechanics ◽  
State Of The Art ◽  
Asymptotic Methods ◽  
Perturbation Series ◽  
Review Article ◽  
Point Of View ◽  
Advantages And Disadvantages ◽  
New Methods ◽  
Local Character

In this review article, we present in some detail new trends in application of asymptotic techniques to mechanical problems. First we consider the various methods which allows for the possibility of extending the perturbation series application space and hence omiting their local character. While applying the asymptotic methods very often the following situation appears: an existence of the asymptotics ε → 0 implies an existence of the asymptotics ε → ∞ (or, in a more general sense, ε → a and ε → b). Therefore, an idea of constructing a single solution valid for a whole interval of parameter ε changes is very attractive. In other words, we discuss a problem of asymptotically equivalent function constructions possessing for ε → a and ε → b a known asymptotic behavior. The defined problems are very important from the point of view of both theoretical and applied sciences. In this work, we review the state-of-the-art, by presenting the existing methods and by pointing out their advantages and disadvantages, as well as the fields of their applications. In addition, some new methods are also proposed. The methods are demonstrated on a wide variety of static and dynamic solid mechanics problems and some others involving fluid mechanics. This review article contains 340 references.

Master curves for the response of layered systems to compressional seismic waves

1967 ◽  
Vol 57 (3) ◽  
pp. 515-543 ◽  
Author(s):  
Luis M. Fernandez
Keyword(s):  
Seismic Waves ◽  
Transfer Functions ◽  
Point Of View ◽  
Time Lags ◽  
Master Curves ◽  
Layer System ◽  
Crust And Upper Mantle ◽  
Response Characteristics ◽  
Families Of Curves ◽  
Short Time

abstract The layers of the earth's crust act as a filter with respect to seimic energy arriving at a given station. Consequently the motion recorded at the surface depends not only on the frequency content of the source and on the response characteristics of the recording instrument, but also on the elastic parameters and thicknesses of the transmitting layers. This latter dependence is the basis for a method of investigating the structure of the crust and upper mantle. To facilitate this investigation a set of master curves for the transfer functions of the vertical and horizontal component of longitudinal waves and their ratios is presented. The calculation of these curves is in terms of a dimensionless parameter. This calculation allows one to group the curves corresponding to different crustal models into families of curves. The characteristics of these curves are discussed from the point of view of their “periodicity” in the frequency domain and of their amplitude in order to investigate the influence of the layer parameters. Considerations, either of constructive interference or of Fourier analysis of a pulse multiply reflected within the layer system, reveal that the amplitudes of the transfer curves depend on the velocity contrasts at the interfaces of the system. The “periodicity” or spacing of the peaks depends on the time lags between the first arrivals and the arrivals of the different reverberations. Closely spaced fluctuations correspond to large-time lags, and widely spaced fluctuations to short-time lags.

scholarly journals Use And Comparison of Topis And Electre Methods In Personnel Selection

2018 ◽  
Vol 22 ◽  
pp. 01021 ◽  
Author(s):  
Bilgin Şenel ◽  
Mine Şenel ◽  
Gizem Aydemir
Keyword(s):  
Personnel Selection ◽  
Selection Process ◽  
Minimum Cost ◽  
Point Of View ◽  
Blue Collar Worker ◽  
Automotive Company ◽  
Worker Selection ◽  
Electre Method ◽  
The Right ◽  
Short Time

One of the most important function of human resources is personnel selection process. This process should be done professionally, in a short time and with minimum cost. After personnel selection process, performance of the hired person is very important for the permanence and success of the company. From this point of view, the aim of this study is to select a personnel among the candidates efficiently, with minimum cost and within a short time in one of the leading companies of Turkey in automotive sector. In order to select the right personnel all criterias which has great impact on blue collar worker selection was decided and these criterias are weighted. From the candidate pool of automotive company, appropriate candidates were selected by using TOPSIS AND ELECTRE method which are multi-criteria decision making methods

Some Possible Lines of Development in Aircraft Engines

1921 ◽  
Vol 25 (123) ◽  
pp. 130-165
Keyword(s):  
Present State ◽  
State Of The Art ◽  
Point Of View ◽  
Aircraft Engines ◽  
Aero Engine ◽  
The Ideal

In the following paper the writer's aim is to indicate certain possible lines of development and research which his own investigations and preliminary experiments have shown to be at least worthy of serious consideration.If we review the present state of the art we find the position to be substantially as follows :—From a thermodynamic point of view the performance of the modern aero engine has approached so nearly to the ideal obtainable from the cycle on which it operates that there is little scope for improvement.

scholarly journals THE VALUATION OF SELF-FUNDING INSTALMENT WARRANTS

2017 ◽  
Vol 20 (04) ◽  
pp. 1750025
Author(s):  
J. N. DEWYNNE ◽  
N. EL-HASSAN
Keyword(s):  
Monte Carlo ◽  
Finite Difference ◽  
Multiple Scales ◽  
Correction Term ◽  
Approximate Solutions ◽  
Fair Value ◽  
Small Time ◽  
Point Of View ◽  
First Order Correction ◽  
Short Time

We present two models for the fair value of a self-funding instalment warrant. In both models we assume the underlying stock process follows a geometric Brownian motion. In the first model, we assume that the underlying stock pays a continuous dividend yield and in the second we assume that it pays a series of discrete dividend yields. We show that both models admit similarity reductions and use these to obtain simple finite-difference and Monte Carlo solutions. We use the method of multiple scales to connect these two models and establish the first-order correction term to be applied to the first model in order to obtain the second, thereby establishing that the former model is justified when many dividends are paid during the life of the warrant. Further, we show that the functional form of this correction may be expressed in terms of the hedging parameters for the first model and is, from this point of view, independent of the particular payoff in the first model. In two appendices we present approximate solutions for the first model which are valid in the small volatility and the short time-to-expiry limits, respectively, by using singular perturbation techniques. The small volatility solutions are used to check our finite-difference solutions and the small time-to-expiry solutions are used as a means of systematically smoothing the payoffs so we may use pathwise sensitivities for our Monte Carlo methods.

Sign in / Sign up

Export Citation Format

Share Document