@inproceedings{HummelBetzoldKandasamyetal.2020, author = {Hummel, S. and Betzold, C. and Kandasamy, K. and Bordin, S. and Harhausen, G. and Dentel, A.}, title = {Experimentelle Validierung von Simulationsmodellen an einem Hardware-in-the-Loop W{\"a}rmepumpenpr{\"u}fstand}, series = {BauSIM 2020}, booktitle = {BauSIM 2020}, doi = {10.3217/978-3-85125-786-1-35}, pages = {299 -- 306}, year = {2020}, abstract = {Basierend auf einer experimentellen Messkampagne werden zwei Simulationsmodelle mit unterschiedlichen Modellierungstiefen, welche die Anlagenkomponenten einer Sole-Wasser-W{\"a}rmepumpe und eines thermischen Energiespeichers abbilden, validiert. Auf Grundlage von Referenzlastprofilen nach VDI 4655 werden Messungen an einem Hardware-in-the-Loop W{\"a}rmepumpenpr{\"u}fstand unter realit{\"a}tsnahen Betriebssituationen und Randbedingungen durchgef{\"u}hrt. Die Validierungsergebnisse zeigen generell eine hohe Genauigkeit des detaillierten Modells gegen{\"u}ber den gemessenen Systemgr{\"o}ßen, aber auch dem vereinfachten Modell, auf. So k{\"o}nnen beispielsweise absolute Fehler im Ladezustand des Speichers von h{\"o}chstens 7,2 Prozentpunkten und von der mittleren Speichertemperatur von 1,4 °C ermittelt werden. Der Einsatz des detaillierten Modells ist f{\"u}r eine zuk{\"u}nftige Entwicklung von Regelungsstrategien zu bevorzugen}, language = {de} } @inproceedings{BetzoldDentelBordin2020, author = {Betzold, C. and Dentel, A. and Bordin, S.}, title = {Entwicklung und Implementierung von Betriebsf{\"u}hrungsstrategien in einem Plusenergiegeb{\"a}ude}, series = {BauSIM 2020}, booktitle = {BauSIM 2020}, doi = {10.3217/978-3-85125-786-1-16}, pages = {144 -- 151}, year = {2020}, abstract = {Der steigende Einsatz erneuerbarer Energien im Wohnungsbau erfordert intelligente L{\"o}sungen, um die lokal erzeugte Energie vor Ort nutzen zu k{\"o}nnen. Das Ziel dieser Forschungsarbeit ist es, optimierte Betriebsf{\"u}hrungsstrategien f{\"u}r modulierende W{\"a}rmepumpen (MWP) mit Speichern zu entwickeln und diese in einem Plusenergie-Reihenhauskomplex anzuwenden, um den Eigenverbrauch der Photovoltaikerzeugung (PV) und die Effizienz zu erh{\"o}hen. Der Betrieb und der Vergleich zweier optimierter Betriebsf{\"u}hrungsstrategien zeigen, dass beide einen {\"a}hnlich hohen PV-Eigenverbrauch erzielen, aber zu unterschiedlichen Arbeitszahlen f{\"u}hren.}, language = {de} } @inproceedings{BayerlHoenigReisteretal.2020, author = {Bayerl, Sebastian P. and H{\"o}nig, Florian and Reister, Joelle and Riedhammer, Korbinian}, title = {Towards automated assessment of stuttering and stuttering therapy}, doi = {10.48550/arXiv.2006.09222}, pages = {10}, year = {2020}, abstract = {Stuttering is a complex speech disorder that can be identified by repetitions, prolongations of sounds, syllables or words and blocks while speaking. Severity assessment is usually done by a speech therapist. While attempts at automated assessment were made, it is rarely used in therapy. Common methods for the assessment of stuttering severity include percent stuttered syllables (\% SS), the average of the three longest stuttering symptoms during a speech task or the recently introduced Speech Efficiency Score (SES). This paper introduces the Speech Control Index (SCI), a new method to evaluate the severity of stuttering. Unlike SES, it can also be used to assess therapy success for fluency shaping. We evaluate both SES and SCI on a new comprehensively labeled dataset containing stuttered German speech of clients prior to, during and after undergoing stuttering therapy. Phone alignments of an automatic speech recognition system are statistically evaluated in relation to their relative position to labeled stuttering events. The results indicate that phone length distributions differ in respect to their position in and around labeled stuttering events.}, language = {en} } @inproceedings{MaiWagnerStreit2020, author = {Mai, Annette and Wagner, Bernhard and Streit, Fabian}, title = {Elimination of Current Harmonics in Electrical Machines with Iterative Learning Control}, volume = {2020}, doi = {10.1109/EDPC51184.2020.9388177}, pages = {1-5}, year = {2020}, abstract = {The magnitude of current harmonics depends on the design of an electrical machine. By suppressing these harmonics noise can be reduced and efficiency improved. Iterative Learning Control (ILC) has proven effective in reducing harmonics. One of the challenges of working with ILC is operation at varying speeds. Variable speeds are particularly important for applications like automotive drives. The ILC period length changes during the learning process at varying speeds. Due to fixed sample rates, the number of values processed by the ILC varies with motor speed.This paper proposes a method to solve this problem and uses ILC at varying speeds. The ILC used to eliminate the harmonics is based on the inverse system. The usage of a two-dimensional memory array is proposed. This data structure holds rows for specific speeds between which interpolation is performed, enabling the elimination of errors which are periodically cyclic to one electrical rotation. This includes the reduction of the motor current harmonics. To verify the presented method a permanent magnet synchronous motor with distinctive 5th and 7th harmonics is used. In real-time implementations, limitations of memory and computational capacity occur.}, language = {en} } @inproceedings{VasquezCorreaBockletOrozcoArroyaveetal.2020, author = {Vasquez-Correa, J. C. and Bocklet, T. and Orozco-Arroyave, J. R. and N{\"o}th, E.}, title = {Comparison of User Models Based on GMM-UBM and I-Vectors for Speech, Handwriting, and Gait Assessment of Parkinson's Disease Patients}, series = {ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)}, booktitle = {ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)}, publisher = {IEEE}, isbn = {978-1-5090-6631-5}, doi = {10.1109/icassp40776.2020.9054348}, pages = {6544 -- 6548}, year = {2020}, abstract = {Parkinson's disease is a neurodegenerative disorder characterized by the presence of different motor impairments. Information from speech, handwriting, and gait signals have been considered to evaluate the neurological state of the patients. On the other hand, user models based on Gaussian mixture models - universal background models (GMMUBM) and i-vectors are considered the state-of-the-art in biometric applications like speaker verification because they are able to model specific speaker traits. This study introduces the use of GMM-UBM and i-vectors to evaluate the neurological state of Parkinson's patients using information from speech, handwriting, and gait. The results show the importance of different feature sets from each type of signal in the assessment of the neurological state of the patients.}, language = {en} } @article{SchusterAriasVergaraMuellerHoerneretal.2020, author = {Schuster, Maria and Arias-Vergara, Tom{\´a}s and M{\"u}ller-H{\"o}rner, Rainer and Winterholler, Cordula and Bocklet, Tobias}, title = {"Verstehen mich mit der Maske eigentlich noch alle?"}, series = {Geriatrie-Report}, volume = {15}, journal = {Geriatrie-Report}, number = {3}, publisher = {Springer Science and Business Media LLC}, issn = {1862-5363}, doi = {10.1007/s42090-020-0627-3}, pages = {8 -- 9}, year = {2020}, language = {de} } @inproceedings{BayerlFrasettoJauernigetal.2020, author = {Bayerl, Sebastian P. and Frasetto, Tommaso and Jauernig, Patrick and Riedhammer, Korbinian and Sadeghi, Ahmad-Reza and Schneider, Thomas and Stapf, Emmanuel and Weinert, Christian}, title = {Offline Model Guard}, doi = {10.48550/arXiv.2007.02351}, pages = {6}, year = {2020}, abstract = {Performing machine learning tasks in mobile applications yields a challenging conflict of interest: highly sensitive client information (e.g., speech data) should remain private while also the intellectual property of service providers (e.g., model parameters) must be protected. Cryptographic techniques offer secure solutions for this, but have an unacceptable overhead and moreover require frequent network interaction. In this work, we design a practically efficient hardware-based solution. Specifically, we build Offline Model Guard (OMG) to enable privacy-preserving machine learning on the predominant mobile computing platform ARM - even in offline scenarios. By leveraging a trusted execution environment for strict hardware-enforced isolation from other system components, OMG guarantees privacy of client data, secrecy of provided models, and integrity of processing algorithms. Our prototype implementation on an ARM HiKey 960 development board performs privacy-preserving keyword recognition using TensorFlow Lite for Microcontrollers in real time.}, language = {en} } @inproceedings{ŁopatkaBocklet2020, author = {Łopatka, Kuba and Bocklet, Tobias}, title = {State Sequence Pooling Training of Acoustic Models for Keyword Spotting}, series = {Interspeech 2020}, booktitle = {Interspeech 2020}, publisher = {ISCA}, address = {ISCA}, issn = {2958-1796}, doi = {10.21437/interspeech.2020-2722}, pages = {4338 -- 4342}, year = {2020}, abstract = {We propose a new training method to improve HMM-based keyword spotting. The loss function is based on a score computed with the keyword/filler model from the entire input sequence. It is equivalent to max/attention pooling but is based on prior acoustic knowledge. We also employ a multi-task learning setup by predicting both LVCSR and keyword posteriors. We compare our model to a baseline trained on frame-wise cross entropy, with and without per-class weighting. We employ a low-footprint TDNN for acoustic modeling. The proposed training yields significant and consistent improvement over the baseline in adverse noise conditions. The FRR on cafeteria noise is reduced from 13.07\% to 5.28\% at 9 dB SNR and from 37.44\% to 6.78\% at 5 dB SNR. We obtain these results with only 600 unique training keyword samples. The training method is independent of the frontend and acoustic model topology.}, language = {en} } @inproceedings{GeorgesHuangBocklet2020, author = {Georges, Munir and Huang, Jonathan and Bocklet, Tobias}, title = {Compact Speaker Embedding: lrx-Vector}, series = {Interspeech 2020}, booktitle = {Interspeech 2020}, publisher = {ISCA}, address = {ISCA}, issn = {2958-1796}, doi = {10.21437/interspeech.2020-2106}, pages = {3236 -- 3240}, year = {2020}, abstract = {Deep neural networks (DNN) have recently been widely used in speaker recognition systems, achieving state-of-the-art performance on various benchmarks. The x-vector architecture is especially popular in this research community, due to its excellent performance and manageable computational complexity. In this paper, we present the lrx-vector system, which is the low-rank factorized version of the x-vector embedding network. The primary objective of this topology is to further reduce the memory requirement of the speaker recognition system. We discuss the deployment of knowledge distillation for training the lrx-vector system and compare against low-rank factorization with SVD. On the VOiCES 2019 far-field corpus we were able to reduce the weights by 28\% compared to the full-rank x-vector system while keeping the recognition rate constant (1.83\% EER).}, language = {en} } @inproceedings{ChenHuangBocklet2020, author = {Chen, Wenda and Huang, Jonathan and Bocklet, Tobias}, title = {Length- and Noise-Aware Training Techniques for Short-Utterance Speaker Recognition}, series = {Interspeech 2020}, booktitle = {Interspeech 2020}, publisher = {ISCA}, address = {ISCA}, issn = {2958-1796}, doi = {10.21437/interspeech.2020-2872}, pages = {3835 -- 3839}, year = {2020}, abstract = {Speaker recognition performance has been greatly improved with the emergence of deep learning. Deep neural networks show the capacity to effectively deal with impacts of noise and reverberation, making them attractive to far-field speaker recognition systems. The x-vector framework is a popular choice for generating speaker embeddings in recent literature due to its robust training mechanism and excellent performance in various test sets. In this paper, we start with early work on including invariant representation learning (IRL) to the loss function and modify the approach with centroid alignment (CA) and length variability cost (LVC) techniques to further improve robustness in noisy, far-field applications. This work mainly focuses on improvements for short-duration test utterances (1-8s). We also present improved results on long-duration tasks. In addition, this work discusses a novel self-attention mechanism. On the VOiCES far-field corpus, the combination of the proposed techniques achieves relative improvements of 7.0\% for extremely short and 8.2\% for full-duration test utterances on equal error rate (EER) over our baseline system.}, language = {en} }