A new research project from the Spanish Research Agency has been (provisionally) granted to AUDIAS. The project is entitled “DeepMUSE: Multi-task and Semi-supervised Deep Learning for Speech and Audio Processing”. It starts in 2022 and will end by the end of 2024. The project builds on the success of a consecutive series of six R&D projects since 2003, and aims at exploring and developing new algorithms focusing on two main research directions: semi-supervised or unsupervised training, and Multi-task or multi-level deep learning models.