[sdiy] ot: rotating speaker simulation or stupid approach

[HOSHUYAMA]

Hello Martin,

Theory of array signal processing and beamforming will be useful for you
to analyze what you want to do (beam scan using loudspeaker array).

Search with keywords such as "beamforming", "beamformer", and "Spatial
aliasing"
would give you a lot of information you need.

I think you have to consider at least two things. (I hope you understand 
what I mean.)

* Spatial aliasing
  Loudspeaker array can be considered as spatial sampling, where the
distance
  of the loudspeakers is impotent.  When the distance is longer than the
  half wavelength of the highest frequency component (1 cm or less), its 
  directivity has multiple peaks called "grating lobes" or "spatial
aliasing".
  Its directivity (beampattern) is like comb filter.  Of course, you
have to 
  take into account the directivity of each loudspeaker.  If each
loudspeaker 
  has sufficiently sharp/narrow directivity at the frequency, aliasing
could be 
  negligible.  However, too sharp directivity is not good to form
interpolated 
  beams in between the loudspeakers.  The directivity of loudspeakers
depends
  on the frequency/wavelength.  Designing beamfomers for wideband
signals like 
  speech is a very difficult problem.  

* Directivity modulation
  By time varying interpolation, the directivity of the loudspeaker
array
  also varies, which is not so simple as that of rotating horn speakers
or
  reflectors.  When you use many loudspeakers, this effect will not be
so 
  important, though.

>From musical point of view, these artifacts could be new effects. ;-)

Sam Hoshuyama
Kawasaki, Japan

Czech Martin wrote:
> 
> Why would the doppler effect be lost?
> 
> Imagine the following: 100 speakers in a single line, every cm a speaker.
> A "sequencer" with interpolation and VCA will make the active of those
> speakers move forth and back. What is the difference of that arraw
> to a single speaker on a model railway track waggon, moving
> foth and back?
> 
> The question is: how many virtual speakers do I need in order to avoid
> cancelation effects when doing the interpolation from one
> speaker to the next.
> 
> m.c.