[sdiy] Interesting demixing technique based on video cues.

[cheater00 cheater00]

yeah, I can easily imagine an improved version of this working to select
one face out of a thousand in a crowd. What defense applications were you
thinking of?

On Sat, 7 Jul 2018, 18:55 Richie Burnett, <rburnett at richieburnett.co.uk>
wrote:

> Incredible! I guess it's no longer "blind source separation" if the
> algorithm can actually see the sources?!
>
> I'm sure defence and security agencies would be interested in this
> technology if it works as well as portrayed.
>
> -Richie,
>
> Sent from my Xperia SP on O2
>
> ---- cheater00 cheater00 wrote ----
>
> You can take a video of multiple instruments and select one to hear just
> that one.
> http://sound-of-pixels.csail.mit.edu/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://synth-diy.org/pipermail/synth-diy/attachments/20180707/ea1b2c6a/attachment.htm>