Refine
Year of publication
- 2013 (77) (remove)
Document Type
- Conference Proceeding (27)
- Contribution to a Periodical (22)
- Article (peer reviewed) (9)
- Part of a Book (8)
- Report (6)
- Other (3)
- Book (2)
Is part of the Bibliography
- no (77)
Keywords
- Holzbau (12)
- Schallmesstechnik (10)
- Trockenbau (4)
- Cement (3)
- Construction Chemistry (3)
- Helium nanodroplets (3)
- CFT (2)
- Doppelhäuser (2)
- FMEA (2)
- Gebäudetrennwand (2)
Institute
- Fakultät für Angewandte Natur- und Geisteswissenschaften (16)
- Fakultät für Angewandte Gesundheits- und Sozialwissenschaften (14)
- Fakultät für Holztechnik und Bau (14)
- Fakultät für Informatik (9)
- Fakultät für Chemische Technologie und Wirtschaft (7)
- Fakultät für Betriebswirtschaft (5)
- Fakultät für Ingenieurwissenschaften (4)
- Fakultät für Innenarchitektur, Architektur und Design (2)
- Forschung und Entwicklung (2)
- Fakultät für Sozialwissenschaften (1)
In safety analysis for safety-critical embedded systems, methods such as FMEA and fault trees (FT) are strongly established in practice. However, the current shift towards model-based development has resulted in various new safety analysis methods, such as Component Integrated Fault Trees (CFT). Industry demands to know the benefits of these new methods. To compare CFT to FT, we conducted a controlled experiment in which 18 participants from industry and academia had to apply each method to safety modeling tasks from the avionics domain.
Although the analysis of the solutions showed that the use of CFT did not yield a significantly different number of correct or incorrect solutions, the participants subjectively rated the modeling capacities of CFT significantly higher in terms of model consistency, clarity, and maintainability. The results are promising for the potential of CFT as a model-based approach.
We describe a state-of-the-art large vocabulary continuous speech recognition (LVCSR) and keyword search (KWS) system trained on roughly 70 hours of conversational telephone speech. Using the Kaldi speech recognition toolkit, we investigate several aspects: for the acoustic front-end, we analyze the use of mel-frequency cepstral coefficients (MFCC), pitch and probability-of-voicing (PoV), and deep neural network (DNN) bottleneck (BN) features, as well as their feature-level combination ("tandem"). For the acousticphonetic decision tree, we explore different hidden Markov model (HMM) topologies for the glottalization phoneme /?/ to model its typically short duration. For the acoustic model, we compare regular continuous HMM with a sort of multi-codebook subspace Gaussian mixture model (SGMM) that lead to an overall best word error rate (WER) of 58.7% and 56.3%, respectively. The KWS is implemented as a word lattice search, and is augmented by a syllable lattice back-up search to capture out-of-vocabulary keywords as well as misrecognized lexical surface forms due to ambiguous prefix and hyphenation rules.
Alleinerziehende
(2013)
Cleft Lip and Palate (CLP) is among the most frequent congenital abnormalities. The impaired facial development affects the articulation, with different phonemes being impacted inhomogeneously among different patients. This work focuses on automatic phoneme analysis of children with CLP for a detailed diagnosis and therapy control. In clinical routine, the state-of-the-art evaluation is based on perceptual evaluations. Perceptual ratings act as ground-truth throughout this work, with the goal to build an automatic system that is as reliable as humans. We propose two different automatic systems focusing on modeling the articulatory space of a speaker: one system models a speaker by a GMM, the other system employs a speech recognition system and estimates fMLLR matrices for each speaker. SVR is then used to predict the perceptual ratings. We show that the fMLLR-based system is able to achieve automatic phoneme evaluation results that are in the same range as perceptual inter-rater-agreements.
The Bode catchment (Germany) shows strong land use gradients from forested parts of the National Park (23 % of total land cover) to agricultural (70 %) and urbanised areas (7 %). It is part of the Terrestrial Environmental Observatories of the German Helmholtz association. We performed a biogeochemical analysis of the entire river network. Surface water was sampled at 21 headwaters and at ten downstream sites, before (in early spring) and during the growing season (in late summer). Many parameters showed lower concentrations in headwaters than in downstream reaches, among them nutrients (ammonium, nitrate and phosphorus), dissolved copper and seston dry mass. Nitrate and phosphorus concentrations were positively related to the proportion of agricultural area within the catchment. Punctual anthropogenic loads affected some parameters such as chloride and arsenic. Chlorophyll a concentration and total phosphorus in surface waters were positively related. The concentration of dissolved organic carbon (DOC) was higher in summer than in spring, whereas the molecular size of DOC was lower in summer. The specific UV absorption at 254 nm, indicating the content of humic substances, was higher in headwaters than in downstream reaches and was positively related to the proportion of forest within the catchment. CO2 oversaturation of the water was higher downstream compared with headwaters and was higher in summer than in spring. It was correlated negatively with oxygen saturation and positively with DOC concentration but negatively with DOC quality (molecular size and humic content). A principle component analysis clearly separated the effects of site (44 %) and season (15 %), demonstrating the strong effect of land use on biogeochemical parameters.