Projects
Aktuelle Projekte
Lifespan AI
Duration and funding organization: since 2023, Deutsche Forschungsgemeinschaft(DFG)
The massive influx of health data in combination with recent advancements in artificial intelligence holds great potential to transform the health sciences. In LifeSpanAI, we envision to advance AI methods and tools to model, predict, and explain health outcomes from multi-dimensional data that span the entire life.
more
SilPara
The aim of SilPara is to lay the foundations of Silent Paralinguistics (SP), i.e., first, to infer speaker states and traits from speech-related biosignals during silent speech production and second, to use this inferred paralinguistic information for a more natural spoken conversation. Finally, we will study the case of speech features involved in politeness and frustration to improve our understanding of their role in communication
Hate speech
Duration and funding organization: since 2021. Deutsche Forschungsgemeinschaft (DFG) (Gepris) and National Science Centre, Poland
Hate speech, i.e., verbal aggression targeting members of various social groups, is widespread in everyday life. It has detrimental consequences for everybody exposed to it. Through our research, we aim to understand both, the underlying properties of hate speech, and its impact on listeners.
more
Smarthelm
In the SmartHelm project, an intelligent helmet is being developed to support courier drivers on the last mile. The helmet has augmented-reality glasses such that the drivers can carry out their tasks with computer assistance, but in a "hands-free" manner. It uses electroencephalography (EEG) and eye movements to detect the person's attention state such that the user interface can be automatically adjusted. The SmartHelm also has multilingual speech recognition such that that users can send voice commands to the system.
more
Adspeed
In the research project ADSPEED our team is working on the decoding of imagined speech in real-time. The goal is to use modern machine learning methods to reconstruct audible speech from invasively measured brain signals and thus explore fundamental questions for a natural and practical communication device that operates in real time and restores natural communication for people with severe speech impairments.
more
Leibniz Wissenschafts-Campus Digital Public Health
Duration and funding organization: 1. Phase 2019-2023 and 2. Phase 2023-2027, Leibniz-Gemeinschaft/Bund/Land
The campus will focus on the digitization of health with smart technologies and how to use data for the benefit of all, while maintaining privacy and data security and bridging the digital divide between populations.
more
KD²School
Duration and funding organization: Until 03/2026 (1st phase), Deutsche Forschungsgemeinschaft(DFG)
The KD2School is a DFG-funded research training group with the topic "Design of adaptive systems for economic decisions". A central aspect of the Research Training Group is its interdisciplinary approach, which combines research from the fields of computer science, economics, and psychology. The goal is to investigate the human decision-making process via the interaction of personal, task-related and environmental factors. With the help of LabLinking, cross-site studies are created that link working groups at the University of Bremen, the Karlsruhe Institute of Technology and the University of Giessen.
MyVoice
In the MyVoice project, a speech prosthesis is being developed that uses the activity of facial muscles during speech production. Users of the system can "speak silently" by moving their facial muscles without producing audible sounds. The system processes the recorded muscle activity using machine learning models to synthesize the acoustic speech signal.
ETAP
Professional long-term care has already been under increased pressure for several years. Due to shortage in labor and hence increased workload, a high health burden is on employees in the care sector and documents the precarious situation. With ETAP we are developing an artificial intelligence (AI)-based movement monitoring system for elderly people in need of care and implementing it in outpatient and inpatient long-term care. Partial automation for fall detection and for recording functional mobility status, partial automation of the associated documentation are the major goals of the project. For this purpose, parts of the nursing documentation related to fall events, and functional mobility as well as positioning protocols are filled AI-supported and automatically. We are developing and investigating AI solutions for automatic fall detection, monitoring movements of residents in care facilities using depth camera images with the help of modern machine learning algorithms.
more
CCSR
It has been consistently found that humans tend to copy aspects of other people’s behavior (ie “behavioral matching”, “behavioral or emotional contagion”, “automatic imitation” “chameleon effect”) in social situations. Such copying of behavior is thought to be important for the outcome of social interactions. Our approach aims to understand the mechanisms underlying such “social resonance”.
more
Nako+Ilse
Duration and funding organization: 2021–12/2024 U Bremen Research Alliance/AI Center for Health Care, financially supported by the Federal State of Bremen
In NAKO+ILSE we team up with Fraunhofer MEVIS to investigate a multi-modal approach for early dementia detection. To this end, we merge conversational speech, as in our previous work with the ILSE study, with brain imaging. To achieve this, we collect a core overlap dataset of autobiographical interviews and MRI data from 500 participants of the German National Cohort (NAKO). Based on this core overlap, our aim is to fuse the NAKO and ILSE populations into a shared dataset, allowing for enhanced early dementia detection
more
Graduiertenkolleg "Empowering Digital Media"
Laufzeit und Fördergeber: Seit 2017, Klaus Tschira Stiftung
Der Schwerpunkt des interdisziplinären Graduiertenkollegs liegt auf Untersuchungen, wie die Mensch-Machine-Interaktion gestaltet werden kann, damit resultierende Systeme sowohl für Individuen als auch für die Gesellschaft einen gesamtheitlich und nachhaltig positiven Effekt haben.
more
GlobalPhone
Duration and funding organization: since 1995, mixed Funding
As part of the GlobalPhone project, speech has been recorded in many countries around the world since 1995. For each language, about 100 native speakers read out newspaper texts. The texts and speech are used to develop pronunciation dictionaries, and large language models are also provided based on extensive text material. GlobalPhone now covers 25 languages, including the top 10 most commonly spoken languages.