On 2011-09-01, [email protected] wrote:
Sure but, as you know, acoustic analysis has long worked in the 40-50 dB SNR territory, whereas for "audio" (i.e. playback of recorded performances) we have gotten used to 80-100 dB -- two very different domains.
But also do note that they're doing cross-modal analysis and active beamforming based on that. So in theory they might well get into the audio range, at least with distant point-like sources, such as speech. That sort of "active decoding" appears to be their main aim as well.
There are very interesting parallels to this in fMRI research, where they try to correlate MRI sequences and/or change-highlighting MRI stills (very high dimensional imaging data) to functional stimuli. I only recently became aware of this to-be modern classic ( http://www.wired.com/wiredscience/2009/09/fmrisalmon/ ), but that's a highly nontrivial problem in statistical correction techniques, and rather interesting if really applied to spatial AV as well.
-- Sampo Syreeni, aka decoy - [email protected], http://decoy.iki.fi/front +358-50-5756111, 025E D175 ABE5 027C 9494 EEB0 E090 8BA9 0509 85C2 _______________________________________________ Sursound mailing list [email protected] https://mail.music.vt.edu/mailman/listinfo/sursound
