@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 = {https://doi.org/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 = {https://doi.org/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 = {https://doi.org/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} } @inproceedings{VasquezBockletOrozcoetal.2020, author = {Vasquez, Juan and Bocklet, Tobias and Orozco, JuanRafael and N{\"o}th, Elmar}, title = {Comparison of user models based on GMM-UBM and i-vectors for speech, handwriting, and gait assessment of Parkinson's disease patients}, doi = {10.48550/arXiv.2002.05412}, 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 (GMM-UBM) 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, Tomas and M{\"u}ller-H{\"o}rner, Rainer and Winterholler, Cordula and Bocklet, Tobias}, title = {Verstehen mich mit der Maske noch alle?}, series = {MMW - Fortschritte der Medizin}, volume = {162}, journal = {MMW - Fortschritte der Medizin}, number = {14}, doi = {https://doi.org/10.1007/s15006-020-0749-4}, pages = {42 -- 44}, year = {2020}, abstract = {In Arztpraxen nimmt der Infektionsschutz mittels Mund-Nasen-Maske momentan eine zentrale Rolle ein. Wenig nachgedacht wird dabei {\"u}ber die Folgen f{\"u}r die Kommunikation. Gerade schwerh{\"o}rigen Patienten bereiten ged{\"a}mpfte Stimmen und abgedeckte Lippen Probleme.}, language = {de} } @inproceedings{LopatkaBocklet2020, author = {Lopatka, Kuba and Bocklet, Tobias}, title = {State Sequence Pooling Training of Acoustic Models for Keyword Spotting}, series = {Proceedings Interspeech 2020}, booktitle = {Proceedings Interspeech 2020}, 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}, doi = {10.48550/arXiv.2008.05011}, pages = {5}, 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} } @article{SchkutowScholleFuhrbergetal.2020, author = {Schkutow, Andreas and Scholle, Karsten and Fuhrberg, Peter and Frick, Thomas}, title = {Scanning techniques for optimized damage tolerance in quasi-simultaneous laser transmission welding of plastics}, series = {Procedia CIRP}, volume = {94}, journal = {Procedia CIRP}, publisher = {Elsevier BV}, issn = {2212-8271}, doi = {https://doi.org/10.1016/j.procir.2020.09.120}, pages = {697 -- 701}, year = {2020}, abstract = {Fiber lasers are increasingly replacing the commonly used diode lasers in laser transmission welding of plastics for challenging applications due to their superior beam quality and the ability to use long working distances and small spot sizes. At the same time, these smaller spot sizes increase the risk of thermal degradation. In this work, we investigate different scanning strategies that allow controlling the weld seam widths. These strategies are characterized regarding the demands on the performance of the scanner system, the risk of thermal degradation and the mechanical properties of the resulting welds.}, language = {en} } @article{EbingerOmondi2020, author = {Ebinger, Frank and Omondi, Bramwel}, title = {Leveraging Digital Approaches for Transparency in Sustainable Supply Chains}, series = {Sustainability}, volume = {12}, journal = {Sustainability}, number = {15}, publisher = {MDPI AG}, issn = {2071-1050}, doi = {https://doi.org/10.3390/su12156129}, year = {2020}, abstract = {With the growing global interdependence of companies, their scope of responsibility for the environmental, social, and human rights impacts associated with their activities is also growing. In this context, companies are increasingly held accountable for social and ecological issues that lie within the sphere of action of their suppliers and sub-suppliers. They are thus faced with the challenge of meeting these demands for transparency, traceability, and compliance with standards in their Supply Chains (SCs). Based on the theoretical framework of Sustainable Supply Chain Transparency (SSCT) in Sustainable Supply Chain Management (SSCM), this conceptual article aims at initiating the discussion on digitalization in SSCM. Therefore, a heuristical research framework, based on relevant fields in the management of sustainability oriented transparency (governance, cooperation and partner selection, traceability/tracking, and strategic and operational risk assessment) is developed. In relation to these fields, consequently, data-driven digital approaches are identified to which potentials for SSCT and control can be attributed. This initial analysis of existing digital approaches already shows that the market is developing dynamically, but is driven more by individual initiatives. In many cases, the approaches used so far are still in the trial phase or offer only limited solutions. Therefore, this paper contributes by giving an overview of the current application of the digitalization approaches in SSCM pinning our discussion on SSCT dimensions.}, language = {en} } @article{DametewBeshahEbinger2020, author = {Dametew, Alie Wube and Beshah, Birehanu and Ebinger, Frank}, title = {The challenges and practice of metal industries into global supply chain integration}, series = {Cogent Engineering}, volume = {7}, journal = {Cogent Engineering}, number = {1}, editor = {Zhou, Zude}, publisher = {Informa UK Limited}, issn = {2331-1916}, doi = {https://doi.org/10.1080/23311916.2020.1762523}, year = {2020}, abstract = {The objective of this study was to investigate the current research practices and existing progress research published on supply chain management and supply chain integration in an attempt to provide a research agenda for future investigations. Mainly the secondary data analysis and methods employed for the investigations. Particularly a systematic literature review analysis from 115 papers published from 16 important journals in the supply chain, logistics, Industrial management \& data systems, supply chain management: an international journal and operations management during the period 1997 to 2018 was employed. This paper has identified relevant supply chain integration research studies that have contributed to the development and growth of manufacturing industries performance and competitiveness to the SCI and SCM fields. The analysis of the relevant articles gaps and miss-concepts in the previous literature provides to reduce universality and insufficient ordinary view, confusion of supply chain integrations were investigated. The gaps and investigations were done based on their methodological porches, objectives, analysis and research finding. Mainly the gaps, miss concepts and challenges concepts in the previous literature were identified and grouped in five issues. Moreover, the gaps and recommendations for future research are consolidated into a research agenda that provides practitioners to evaluate existing SCI \& SCM issues and seek ways to develop a conceptual farm work on SCI of manufacturing industries to global supply chains.}, language = {en} }