On 25 May 2011, at 10:06, Dave Malham wrote: > On 24/05/2011 20:00, [email protected] wrote: >> <snip> >> >> I should mention that interpolation of HRTF is not the only possible >> technique; you can use for example a virtual loudspeaker array... >> > This is certainly the way that the Lake DSP system worked that they > demonstrated way back in 1993 (I think it's in the papers for the London VR93 > confence from that year but I don't have my copy of the proceedings hand). > The sounds were recorded in (first order) Ambisonics and the head tracking > drove a rotate/tilt algorithme that fed a decoder to virtual speakers the > signals from which were convolved with fixed hrtf's corresponding to the > speakers' positions that were fixed wrt the head, mixed together and fed to > the headphones.
Lake also licensed a product for use in airline entertainment systems and similar situations where they had 5.1 movie surround tracks that needed to be delivered to headphones. If I understood correctly how that worked was as follows: a) set up a dummy head b) fire an impulse (or sine sweep) at each of the five speaker positions, so you end up with ten IR files. c) now you can process each of the five channels for each ear, then you sum the five channels for each ear d) => binaural recording or they used a sound field type microphone instead of a dummy head, to make B-Format IRs, and then created a B-fourmat intermediate source, which then was decoded into binaural stereo. The materials I read back then were somewhat vague, and allowed for both interpretation. The lack of mention of Ambisonics led me to believe the first variation was used, but word was they used Ambisonics in the process, which would indicate the second process was used. Ronald _______________________________________________ Sursound mailing list [email protected] https://mail.music.vt.edu/mailman/listinfo/sursound
