On Sun, Mar 04, 2012 at 01:28:10PM -0800, Aaron Heller wrote: > I would not say that I've heard no detrimental effects on large arrays > -- I've heard all sorts of problems, but in every case I've been able > to attribute it to other reasons: bad decoder, errors in speaker > positions, feeding 1st-order signals into a HOA decoder, and so forth. > > For example, decoders where the HF/LF balance is set using the > "conservation of total energy" approach from Daniels' thesis, > emphasize HF more and more as the number speakers goes up. I find > this causes the tonal balance on large arrays to be wrong, as well as > creating near-head artifacts (indicative that the sound at left and > right ears is very different). Reducing the HF/LF balance by 2 to 4 > dB fixes this. I use the Stravinsky and Beethoven recordings I posted > to Ambisonia to judge this, as I am quite familiar how they sound when > reproduced correctly.
Still there are objective reasonss why using too many speakers can be expected not to work well. Take for example a ring of 12 speakers reproducing 1st order. Even if the decoder is 1st order in the sense that it accepts only 1st order inputs, such a decoder is in fact a higher order one, one that assumes that those components that the speaker array *could* reproduce but which are not part of the input, are zero. In this cases, the 2nd...5th order components will be missing in the reconstructed sound field. In a 'real' sound field these are present. And when using the right number of speakers they will be present as well, by aliasing. Of course in the latter case they are wrong, field reconstruction fails and that forces us to ove to an 'rE' based decode as a compromise. But that is still much better than not having these components at all. The result of using too many speakers (following from the Fourier- Bessel sum) is that there will be 'silent ring' around the center, with a radius depending on wavelenght - the area in which the missing components dominate the sound field. This is very visible if you plot the magnitude of the resultant field. It looks like 'focussing' on the sweet spot - this is consistent with the near-head effects you mentioned. And of course the frequency dependence will lead to phasing effects. Ciao, -- FA Vor uns liegt ein weites Tal, die Sonne scheint - ein Glitzerstrahl. _______________________________________________ Sursound mailing list [email protected] https://mail.music.vt.edu/mailman/listinfo/sursound
