Exzellenz Chair Li
Machine Listening
The Machine Listening Lab (MLL) represents one of the U Bremen Excellence Chairs projects and aims to integrate the human ability of Selective Listening into machines.
This ability allows humans to focus on a specific sound source while fading out all remaining sound sources. Typically, this is also referred to as "The Cocktail Party Problem". Hearing-impaired individuals often lose this ability or can only perform it to a limited extend, making hearing aids necessary. Even though the technology in hearing aids is quite advanced, it reaches its limits in real Cocktail Party scenes. At MLL, AI-based algorithms are developed which will be integrated into Smart Hearing Aids in the long term in order to allow hearing-impaired individuals to regain the ability of Selective Listening. Methods from the research fields of "Speech Separation" and "Brain Signal Processing" are used to develop neuro-steered algorithms. Besides this long term goal, the MLL investigates corresponding methods and algorithms in different scenarios, such as:
- Speech Separation in multilingual Cocktail Party scenarios
- The impact of speech mode adaptation (for instance whispered speech) on Speech Separation algorithms
- The classification and tracking of selective auditory attention in complex auditory environments using electroencephalography (EEG)
- Speech Separation algorithms steered by EEG signals
Employees at MLL/CSL:
Director: Prof. Dr. Haizhou Li
Host: Prof. Dr.-Ing. Tanja Schultz
Contact person: Marvin Borsdorf, M.Sc.