Hi, The person to really answer this is Geoffrey barton (or perhaps Peter Craven). I'm sure I have something on this somewhere (possible in the stuff that went to the FCC??) but I can't find it at present. I'll have a think about this, but it may have been something to do with improving on the earlier "Hafler" difference system for extracting ambience from stereo recordings and placing it on a pair of rear speakers, which were often set up as an out of phase pair to try and decorrelate the sound.
Dave On 20 July 2012 09:55, Johan Haspeslagh <johan.haspesl...@pandora.be> wrote: > Hello, > > I have a question concerning the conversion of a stereo recording to > ambisonics. In the Wireless World of 1977 articles there are equations for the > conversion of stereo to W,X,Y equivalent signals. The use of j*Diff term > doesn't seem to make sense. As I found no background info on them, maybe > someone in the group has. > > The WW1977 article states: > W=0.71*Sum-0.291*j*Diff > X=0.71*Sum+0.291*j*Diff > Y=0.583*Diff > where Sum=L+R and Diff=L-R. > > I was puzzled with the addition of the j*Diff, so I tried to find some > mathematical support for it. The reasoning I did is as follows: > > 1. As W,X,Y are a representation of a supposed plane wave with amplitude S > coming from a certain angle phi (phi=0 if straight ahead), we need to suppose > a analoguous encoding for stereo. I supposed a the sin-like law of encoding > f.e.: > > (L-R)/(L+R)=sin(phi)/sin(30) if loudspeakers are placed at -30 and 30 degrees. > > 2. As a starting point a naive conversion of the L,R signals (projection of > L,R on X,Y) to ambisonic-like format gives: > W=Sum > X=Sum*cos(30) > Y=Diff*sin(30) > > If we write Sum and Diff in relation to the supposed direction encoding in (1) > as function of S and phi, then > W=S > X= S*cos(30) > Y= S*sin(phi) > > X is independent of the direction, Y correctly encodes the direction neatly. X > is only correct for the center direction (phi=0) but for all other directions > the magnitude of the encoded vector is to large (sqrt(X²+Y²)>W). > > 3. The adding of the j*Diff terms makes af first sight everything even worse, > it makes the magnitude of X and W higher and it also introduces a phase > differences between X,Y and W which is in theory not possible if one encodes a > plane wave directly. > > The only reason I could think of is that by introducing phase differences once > you construct the loudspeaker signals by adding weighted versions of the W, X > and Y signal together these phase differences introduce lower amplitudes (than > adding signals in phase). I did not go through with this calculation yet. > > So if someone have some idea's on it, > > Johan > > > > > > > > > > > -- > -------------- next part -------------- > An HTML attachment was scrubbed... > URL: > <https://mail.music.vt.edu/mailman/private/sursound/attachments/20120720/47cd95a9/attachment.html> > _______________________________________________ > Sursound mailing list > Sursound@music.vt.edu > https://mail.music.vt.edu/mailman/listinfo/sursound -- These are my own views and may or may not be shared by my employer Dave Malham Music Research Centre Department of Music The University of York Heslington York YO10 5DD UK Phone 01904 322448 Fax 01904 322450 'Ambisonics - Component Imaging for Audio' _______________________________________________ Sursound mailing list Sursound@music.vt.edu https://mail.music.vt.edu/mailman/listinfo/sursound