Stream fusion for multi-stream automatic speech recognition

Multi-stream automatic speech recognition (MS-ASR) has been confirmed to boost the recognition performance in noisy conditions. In this system, the generation and the fusion of the streams are the essential parts and need to be designed in such a way to reduce the effect of noise on the final decision. This paper shows how to improve the performance of the MS-ASR by targeting two questions; (1) How many streams are to be combined, and (2) how to combine them. First, we propose a novel approach based on stream reliability to select the number of streams to be fused. Second, a fusion method based on Parallel Hidden Markov Models is introduced. Applying the method on two datasets (TIMIT and RATS) with different noises, we show an improvement of MS-ASR.     

Generic visual analysis for multi- and hyperspectral image data

Multi- and hyperspectral imaging and data analysis has been investigated in the last decades in the context of various fields of application like remote sensing or microscopic spectroscopy. However, recent developments in sensor technology and a growing number of application areas require a more generic view on data analysis, that clearly expands the current, domain-specific approaches. In this context, we address the problem of interactive exploration of multi- and hyperspectral data, consisting of (semi-)automatic data analysis and scientific visualization in a comprehensive fashion. In this paper, we propose an approach that enables a generic interactive exploration and easy segmentation of multi- and hyperspectral data, based on characterizing spectra of an individual dataset, the so-called endmembers. Using the concepts of existing endmember extraction algorithms, we derive a visual analysis system, where the characteristic spectra initially identified serve as input to interactively tailor a problem-specific visual analysis by means of visual exploration. An optional outlier detection improves the robustness of the endmember detection and analysis. An adequate system feedback of the costly unmixing procedure for the spectral data with respect to the current set of endmembers is ensured by a novel technique for progressive unmixing and view update which is applied at user modification. The progressive unmixing is based on an efficient prediction scheme applied to previous unmixing results. We present a detailed evaluation of our system in terms of confocal Raman microscopy, common multispectral imaging and remote sensing.     

The Formation of Calcified Nanospherites during Micropetrosis Represents a Unique Mineralization Mechanism in Aged Human Bone

Osteocytes—the central regulators of bone remodeling—are enclosed in a network of microcavities (lacunae) and nanocanals (canaliculi) pervading the mineralized bone. In a hitherto obscure process related to aging and disease, local plugs in the lacuno‐canalicular network disrupt cellular communication and impede bone homeostasis. By utilizing a suite of high‐resolution imaging and physics‐based techniques, it is shown here that the local plugs develop by accumulation and fusion of calcified nanospherites in lacunae and canaliculi (micropetrosis). Two distinctive nanospherites phenotypes are found to originate from different osteocytic elements. A substantial deviation in the spherites' composition in comparison to mineralized bone further suggests a mineralization process unlike regular bone mineralization. Clearly, mineralization of osteocyte lacunae qualifies as a strong marker for degrading bone material quality in skeletal aging. The understanding of micropetrosis may guide future therapeutics toward preserving osteocyte viability to maintain mechanical competence and fracture resistance of bone in elderly individuals.     

The Effects of Debonding on the Low-Velocity Impact Response of Steel-CFRP Fibre Metal Laminates

The effect of metal-composite debonding on low-velocity impact response, i.e. on contact force–central deflection response, deformation profiles and strains on the free surfaces was studied. We focused on type 2/1 fibre metal laminate specimens made of stainless steel and carbon fibre epoxy layers, and tested them with drop-weight impact and quasi-static indentation loadings. Local strains were measured with strain gauges and full-field strains with a 3-D digital image correlation method. In addition, finite element simulations were performed and the effects of debonding were studied by exploiting cohesive elements. Our results showed that debonding, either the initial debonding or that formed during the loading, lowers the slope of the contact force–central deflection curve during the force increase. The debonding formation during the rebound phase was shown to amplify the rebound of the impact side, i.e. to lower the ultimate post-impact deflection. The free surface strains were studied on the laminate’s lower surface at the area outside the debond damage. In terms of in-plane strains, debonding formation during impact and indentation, as well as the initial debonding, lowered the peripheral strain and resulted in a positive change in the radial strain.     

The mitigation hierarchy upside down – a study of nature protection measures in Danish infrastructure projects

This paper presents a Danish study of mitigation directed at nature protection in environmental impact assessment (EIA) of infrastructure projects. The study is based on a document analysis of EIA reports, a workshop held with EIA professionals, and a study of two cases. The paper takes a point of departure in the mitigation hierarchy as a central conceptual framework, identifies which mitigation measures have been suggested in the EIA reports, and compares this to the mitigation hierarchy. Further, the paper explores the dynamics behind which mitigation measures are chosen and later implemented. The findings point to a discrepancy between the prevention principle embedded in the mitigation hierarchy and the actual EIA practice with increased use of nature compensation. Further, the research reveals significant variation in the design of mitigation measures, e.g. in the level of detail used in describing them and the level of clarity as to aims and actions.     

The use of social media in impact assessment: experiences among national infrastructure developers in Denmark

Social media plays an increasingly prominent role in the interaction between impact assessments and society. Impact assessment is thus becoming part of social networks that are claimed to be able to improve interaction between actors in impact assessment processes. By investigating the use of social media in impact assessment and planning processes of national linear infrastructure development, we explore how social media impinge on participatory practices and how the potential of social media is realized in the current use related to public participation processes. The study focuses on experiences among national developers working in road, rail, electricity, gas, and metro infrastructures in Denmark. The article shows that the current use of social media in public participatory practices is limited to branding and on-way communication and that the increasing use of social media in linear infrastructure planning gives rise to a variety of concerns among developers that are related to especially organizational cultures, perception of the target groups, and prioritization of resources.     

Ultralow Expansion Glass as Material for Advanced Micromechanical Systems

Ultralow expansion (ULE) glasses are of special interest for temperature stabilized systems for example in precision metrology. Nowadays, ULE materials are mainly used in macroscopic and less in micromechanical systems. Reasons for this are a lack of technologies for parallel fabricating high-quality released microstructures with a high accuracy. As a result, there is a high demand in transferring these materials into miniaturized application examples, realistic system modeling, and the investigation of microscopic material properties. Herein, a technological base for fabricating released micromechanical structures and systems with a structure height above 100 μm in ULE 7972 glass is established. Herein, the main fabrication parameters that are important for the system design and contribute thus to the introduction of titanium silicate as material for glass-based micromechanical systems are discussed. To study the mechanical properties in combination with respective simulation models, microcantilevers are used as basic mechanical elements to evaluate technological parameters and other impact factors. The implemented models allow to predict the micromechanical system properties with a deviation of only ±5% and can thus effectively support the micromechanical system design in an early stage of development.     

Liquidus temperatures for primary crystallization of cryolite in molten salt systems of interest for aluminum electrolysis

Temperatures for primary crystallization of Na₃AlF₆ in multicomponent electrolyte systems of interest for the aluminum electrolysis process were determined by thermal analysis. The results are presented as binary and quasibinary diagrams and discussed in view of the literature data. An empirical equation describing liquidus temperatures for primary crystallization of Na₃AlF₆ was derived: $$\begin{gathered} t/(^\circ C) = 1011 + 0.50[AlF_3 ] - 0.13[AIF_3 ] - \frac{{3.45[CaF_2 ]}}{{1 + 0.0173[CaF_2 ]}} \hfill \\ + 0.124[CaF_2 ] \cdot [AlF_3 ] - 0.00542([CaF_2 ] \cdot [AlF_3 ])^{1.5} \hfill \\ - \frac{{7.93[Al_2 O_3 ]}}{{1 + 0.0936[Al_2 O_3 ] - 0.0017[Al_2 O_3 ]^2 - 0.0023[AlF_3 ] \cdot [Al_2 O_3 ]}} \hfill \\ - \frac{{8.90[LiF]}}{{1 + 0.0047[LiF] + 0.0010[AlF3]^2 }} - 3.95[MgF_2 ] - 3.95 \hfill \\ \end{gathered} $$ wheret is the temperature in degree Celsius and the square brackets denote the weight percent of components in the system Na₃AlF₆-AlF₃-CaF₂-Al₂O₃-LiF-MgF₂-KF. The composition limitations are [AlF₃] ≈ [CaF₂] ≈ [LiF] < 20 wt pct, [MgF₂] ≈ [KF] < 5 wt pct, and [A1₂O₃] up to saturation.     

Affective speaker state analysis in the presence of reverberation

Little is known about the impact of reverberation on performance of running speaker state classification systems. This study thus aims to approach the topic by measuring effects on a state-of-the-art engine with consideration of six public room impulse responses for convolution of the affective speech signals of three standard datasets comprising of emotion and interest. Speech data thereby is given by this year’s INTERSPEECH Paralinguistic Challenge corpus TUM AVIC and the frequently used Berlin and eNTERFACE sets. The room impulse responses comprise rooms in private apartments, chapels, a factory hall, and a van. Speaker independent performance after speaker adaptation is investigated. To cope with reverberation, matched condition learning and acoustic space adaptation are considered as efficient means. By that a report is provided on suitability of feature types given the type of impulse response. In the result almost all occurring corruption arising from reverberation can be restored, yet the general impact varies with the type of room or acoustic environment.     

Implementation of a Modelica compiler using JastAdd attribute grammars

We have implemented a compiler for key parts of Modelica, an object-oriented language supporting equation-based modeling and simulation of complex physical systems. The compiler is extensible, to support experiments with emerging tools for physical models. To achieve extensibility, the implementation is done declaratively in JastAdd, a metacompilation system supporting modern attribute grammar mechanisms such as reference attributes and nonterminal attributes.This paper reports on experiences from this implementation. For name and type analyses, we illustrate how declarative design strategies, originally developed for a Java compiler, could be reused to support Modelica’s advanced features of multiple inheritance and structural subtyping. Furthermore, we present new general design strategies for declarative generation of target ASTs from source ASTs. We illustrate how these strategies are used to resolve a generics-like feature of Modelica called modifications, and to support flattening, a fundamental part of Modelica compilation. To validate that the approach is practical, we have compared the execution speed of our compiler to two existing Modelica compilers.