Live data from Hacker News

FlowTSE: Target Speaker Extraction with Flow Matching

Post reply on HN

Print view View original

1–3 of 3 posts

FlowTSE: Target Speaker Extraction with Flow Matching

#1

Post by agold97 » Wed, May 28, 2025, 2:30 PM UTC

FlowTSE: Target Speaker Extraction with Flow Matching
arxiv.org

Re: FlowTSE: Target Speaker Extraction with Flow Matching

#2

Post by juancn » Wed, May 28, 2025, 5:03 PM UTC

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

Re: FlowTSE: Target Speaker Extraction with Flow Matching

#3

Post by max_ » Wed, May 28, 2025, 5:15 PM UTC

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.

Post reply on HN