The REMUS project aims to enhance historical music recordings using advanced generative models. Imagine hearing century-old musical treasures with the clarity and richness of today's high-quality audio - that's our goal. Utilizing state-of-the-art generative models, grounded in deep neural networks, our initiative will not only remove hiss and crackles from old records but will also restore lost high-frequency sounds and repair damage such as scratches and fade-outs on gramophone and vinyl records. This four-year journey will blend technology with cultural heritage, democratizing audio restoration, offering new tools to musicologists, and preserving our musical past for future generations. The REMUS project is more than just a technological feat; it is about forging a bridge between the past and the present, ensuring historical music will resonate with today's audiences.

2025

2029