Description:
Speech recognition systems are becoming more and more advanced and could potentially replace keyboard-based text input in the future. However, composing a text involves more than pressing keys on a keyboard and cognitive processes involved in writing are substantially different from speech production, involving extensive reviewing and editing of the produced text.
The goal of this is thesis to explore the question of how speech recognition systems need to be designed to best support the complex cognitive processes involved in text composition and how we can leverage large generative models for this. The specific topic can be targeted to the interests of the student and the duration of the thesis (Bachelor or Master). In any case, it will involve building systems using generative AI and conducting empirical user studies.
Requirements:
- Very good programming skills (python, web-based)
- Experience with or willingess to learn building LLM-based applications
- Interest in conducting empirical user studies
Links:
- https://journals.sagepub.com/doi/10.1177/0741088312451260
- https://www.researchgate.net/publication/316996736_The_processing_demands_of_writing
- https://pure.mpg.de/rest/items/item_2152024/component/file_2152025/content
- https://www.jstor.org/stable/1746333
Comments are closed.