392264 Project: Familiar Voices as Calming AI Systems for Emotional Support in Elderly Care (Pj) (WiSe 2025/2026)

Emotional well-being in nursing homes is closely linked to feelings of connection and familiarity. This project investigates whether hearing messages in synthetic but familiar voices can provide comfort and a sense of closeness to elderly residents.
The goal is to use modern voice cloning models to replicate the voices of family members and deliver personalized messages in a way that feels warm and familiar. To ensure that the messages truly meet the needs of the recipients, they are rewritten into clear and simple language by a language model before being converted into speech.
The system is intended to complement regular personal visits, not replace them. Its purpose is to provide emotional security and a sense of presence between interactions with family.
Participants will build a small prototype that combines message input, text customization, and speech synthesis. They will then examine how the messages affect emotional well-being using methods such as mood surveys, interviews, physiological measurements, or speech tone analysis.