FlowTSE: Target Speaker Extraction with Flow Matching

Examples can be found here: https://aiola-lab.github.io/flow-tse/

This is brilliant.

It can allow calls be be made clear when people talking in noisy places.

Assuming this can be made to work in real time.