Being hearing impaired has a big impact on someone’s well being. An important problem for hearing impaired people is the disability to correctly localize sound sources. Apart from annoying, this can be dangerous in practical situations (consider the inability to localize sound sources in traffic), and is an important reason why hearing-impaired people are less confident in practical situations, need more assistance and have a worse quality of life. For localization, normal-hearing people rely to a large extend on time differences. However, initial research has shown that very often, hearing impaired people cannot perceive these time differences. This depends on the individual hearing loss and other aging processes. State-of-the-art hearing aid algorithms tend to preserve the acoustic spatial information from the original sound scene. For hearing impaired people this is thus far from effective as, depending on the user’s hearing loss, the preserved spatial cues cannot be heard. With this proposal we will investigate whether it is possible to transform time differences that are inaudible for a specific user, into audible level differences. As the exact frequency range where time differences cannot be heard is user dependent, the envisioned system will be user specific.

Promo and trailer production are expensive and labor-intensive tasks in media industry. Recently, content-based video summarization has found its way to the field of computer vision and deep learning that can greatly reduce the human effort for this task. Most effective methods in this regard are relying on deep learning approaches, i.e., learning effective video representations directly from raw video frames, without hand crafting them for the summarization task. Deep learning frameworks are known to be data hungry, which require an immersive annotated data for the training procedure. Collecting annotated video for training deep networks is yet another labor-intensive task that should be limited for cheap automatic promo production. The common solution to avoid the annotation costs is to use the pretrained network on existing labeled datasets. Nevertheless, the performance of the pre-trained networks deteriorates gracefully when the properties of the test videos drifts away from that of the training videos. In this project, we aim at quantifying/modeling the visual domain disparity for the video summarization task by means of real data that is provided by our private partner (RTL). This can be effectively done by evaluating the existing deep learning approaches, that are trained and tested on public video datasets, directly on RTL video collections. Moreover, the evaluation of video summarization task is hardly objective due to its complex nature (human reasoning is required). We benefit from the expertise of our industrial partner to develop a framework for instrumental measure of the quality of summarized video content.

In this project, a robust and efficient approach for accurate radio positioning and time-transfer is researched, developed, tested and evaluated, through the use of virtual ultra-wideband radio signals. When applied in a terrestrial radio system, it can serve as a backup and complementary to GNSS (Global Navigation Satellite System) in environments with reduced GNSS accuracy. This approach allows for a limited demand for expensive radio frequency spectrum and time resources compared to existing approaches, yet delivers maximum performance and robustness, and through flexible signal design, implementation and integration with current and new generation telecommunications standards, such as 5G, is expected.