Erlangen to become the flagship of Bavarian AI research
Artificial intelligence (AI) offers a whole range of new approaches for signal processing solutions. A new center was founded in 2020 to systematically develop the technology, and, as a flagship project, it has received 13 million euros in funding over five years from the Bavarian Ministry of Economic Affairs, Regional Development and Energy as a flagship project. Under the leadership of Dr. Frederik Nagel, various AI-assisted basic technologies are being developed in four work areas. They are available for consulting and licensing and, in the long term, for customized solutions. Audio signal processing, natural language user interfaces (NLUI), computer vision and data transmission benefit from the use of AI and machine learning and are further expanding our competitive edge in signal processing on an international level.
Audio quality and speech recognition
The participants in a video conference are annoyed. Somebody nearby is constantly typing on their keyboard and bothering everyone with their tip-tapping. “With the methods of artificial intelligence, we will be in a position in the future to alter voice signals so that they can be understood without noise interference and with virtually no loss of audio quality,” says Dr. Nagel, outlining technological progress that will of benefit not only for future participants in online conferences but also in other fields of application. In television, AI-based techniques can clearly distinguish background noises and dialog from each other, so that viewers can individually adjust their respective volumes. In addition, the experts are carrying out research in the field of audio signal processing and voice assistants for secure communication with organizations such as banks and insurance companies, where the data has to remain in Germany.
Image processing and communication
Whereas online customers today must make do with a conventional, two-dimensional picture of e-commerce items, the use of new AI methodology will help provide a 360-degree image containing detailed information about the qualities of the merchandise. “In image processing, we’re expecting new solutions for the efficient generation of virtual scenes from individual pictures,” the head of department announces. “And if you are no longer annoyed by poor network coverage and low robustness of transmission in the future, this will be partly owing to the combination of AI and signal processing in communication systems.” Wireless transmission systems have long been a core area of Fraunhofer research. Their use in shared radio systems for communication and radar sensing can contribute to optimized resource allocation and therefore save valuable energy.