Hacker News Reader

Not even ultrasound, just correlate what everybody hears at the same time?

They might just need one data point, if the playlist curator tells them the exact moment at which the song started.