LAOLA - Interactive therapy support using AI-based mouth, posture and speech recognition, using the example of voice disorder therapy: Motion Analysis & Decision Support

The aim of the LAOLA project is to research a speech therapy app to support the therapy of voice disorders. The app provides interactive training for patients using digital online analysis of visual (analysis of mouth movements, joint point analysis) and auditory aspects (voice recognition). In the course of LAOLA, an extended concept for therapy support will be developed combining online analysis of mouth movements, posture and voice with direct feedback (e.g. posture correction) during video-based exercise instruction. This online feedback replaces the feedback function of the therapist and enables precise exercise execution.

The goal of the sub-project at the University of Lübeck is to design and research the visual sensor analysis for AI-based recognition of mouth movements and body postures, as well as the decision support for the recommendation of training sessions and the online feedback.

Available thesis topics:

Topics Description / Language
Audio-based Voice Impairment Detection DE EN
Depth-Assisted Face Reconstruction: Leveraging RGB and Depth Information for Detailed Face Meshes DE EN
Decision Support for Speech Therapies with Machine Learning DE EN

Project funding / Cooperations

The joint project is funded by the BMBF for the period 2022 - 2025.