BME MDA - a Műegyetem Digitális Archivuma
Not a member yet
47877 research outputs found
Sort by
Improving speech naturalness and nuance using HiFiGAN-Hubert-Soft vocoder: A case study of the Voicebox TTS model
Get PDFText-to-speech (TTS) technology has significantly transformed human-machine interactions, facilitating seamless communication between humans and computers. However, achieving high-quality TTS remains a formidable challenge, especially in synthesizing natural and nuanced speech. In this study, we investigate the potential of HiFiGAN-Hubert-Soft (HHS) vocoder to enhance the performance of TTS models, with a focus on integrating the HHS vocoder into the Voicebox TTS model—a versatile and scalable TTS system developed by Meta AI. Through both subjective (mean opinion score) and objective (audio similarity and visualization metric) evaluations, we illustrate that the HHS vocoder significantly enhances the naturalness and nuance of synthesized speech compared to the baseline HiFiGAN vocoder. This improvement is particularly pronounced in cases where pronunciation variations are subtle or context-dependent. Our findings emphasize the potential of the HHS vocoder in elevating TTS performance and laying the foundation for further advancements in TTS technology
Using fault tolerant design patterns to assure data veracity
No full textData forms the vital asset of many organizations, as the quality of their decisions depends on the quality of their data. Trust in data is, therefore, critical. This paper aims to evaluate different aspects of data quality, examine the existing data veracity characteristics, and propose a methodology to assess the impact of fault-tolerant design patterns on data veracity. Data generated by IoT devices often reveals characteristics such as noise, incompleteness, and imprecision [1], which make it a prime example for data quality assessment. This paper investigates how we can effectively address the attributes and characteristics associated with data veracity by applying fault-tolerant design patterns within the data processing workflow
Contribution of different movement tasks to differential diagnosis of Parkinson’s disease
No full textParkinson's disease is one of the most common neurological diseases, which, according to current knowledge, is incurable. Early detection is essential since, with appropriate therapy and medication, the progress of the disease can be slowed down and the quality of life maintained. The movement tasks described with the acceleration data are part of the intensive research area. This would make recognizing the disease and specific symptoms like tremors, rigidity, and bradykinesia possible. Many research studies focus on the selection of appropriate movement tasks. However, due to the diversity of the studies, no consensus has yet been reached. Therefore, in this research, we examine which of the movement tasks selected from the Unified Parkinson's Disease Rating Scale prove to be the best under the same procedure. Furthermore, we attempted to make a final decision based on the predictions obtained on the movement forms using voting procedures. 37 patients with Parkinson's disease and 47 healthy individuals participated in this study. 3-axial acceleration data from the wrist-mounted sensor was acquired, from which times-series features were determined. Classifications were done by Support Vector Machine. Soft, hard, and SVM-based voting were also explored. The results indicated that the PRONATION task has the highest balanced accuracy (76.2%). Among the voting approaches, the soft achieved the highest improvement (79.8%) compared to the best task. Based on the results, fewer movement tasks can be used to recognize the disease, among which PRONATION is essential. Further voting approaches can improve the performance
Control Tower Approach to Improve Urban Construction Logistics
Get PDFIn many countries the construction industry is responsible for much of freight transport and emissions. In recent years, insight has been gained into the application of various logistics solutions, which have often proven to be successful in reducing construction transport movements and emissions in and around cities. Further steps could be taken to advance information exchange and supply chain collaboration in construction logistics. Supply chain control towers have been deemed to improve supply chain management in individual construction projects, as well as on an area level including multiple projects and supply chains.
The goal of this research has been to lay the foundations and demonstrate the added value of applying the concept of the Construction Logistics Control Tower (CLCT) in urban construction logistics. The research has investigated how CLCTs can help to exchange data and information effectively in order to improve decision making, and reduce negative effects of construction transport.
First a stakeholder analysis looked into the motivations of stakeholders in applying CLCTs. Workshops and surveys provided insight into individual motivations of stakeholders. Preferably stakeholders appreciated insight into the current state of affairs and to use this information to improve logistics efficiency, for a better financial return individually. On the other hand there was appreciation among stakeholders also for CLCTs that are area-oriented and of general use for all.
A case study among selected stakeholders focused on optimizing logistics and applying centralised control by analysing and integrating the logistics of simultaneous construction projects in the centre of Amsterdam. The study demonstrated the added value of centralised logistics management to reduce transport movements for the construction projects, optimize resource usage, and improve the logistics management of the projects
