[LAU] first order ITU 5.1 preset for ambdec or second order stereo panner

[Fons Adriaensen]

Hello Hector,

> Hello (Fons),
> I was wondering if anyone has a ITU 5.1 preset for ambdec. The one
> included is for second order only and I would like  to use the stereo
> panner included in the LADPSA AMB plugins by Fons. A second order
> stereo panner would be great too.

I delayed replying to this because the answers are
not simple.
 
I do have a 1st order ITU 5.1 preset, it's included
at the end of this message. But you should consider
the following. If you are panning individual sources
into first order AMB then the results on a 5.1 rig
could be inferior to what is possible using normal
5.1 panning. It depends a bit on what you expect -
the 1st order images will always be more 'diffuse'
than 'real' 5.1, OTOH, the imaging quality will be
more constant as a function of position, and sound
will appear less to come 'from the speakers'. First
order to ITU 5.1 works well with signals recorded
with an AMB microphone, but for panned material you
want 2nd order.

Regarding the stereo panner, it is equivalent to
two mono panners, just with a different interface.

But panning stereo into 2nd or higher order AMB may
not give the result you expect. It is *not* a correct
mapping from stereo to AMB - this is impossibles as
essential information is lost when encoding to stereo
(if it were possible then higher order AMB could be
encoded without loss into two channels...).

What happens is that you encode two source directions
(the stereo speakers) correctly, and all the rest is
just a linear combination of these. The AMB reproduction
system will create two 'virtual speakers' for your stereo
signal. The higher the order the more accurate this will
be, but the acoustic result is just stereo. A central
sound will be as 'fuzzy' as would be just using normal
stereo reproduction. Even if your AMB rig has a speaker
near to the central position it just won't be used, or
very little, because you didn't pan anything there.
 
How much this matters depends on how you use it.
It will not be worse than stereo, but not better
either while with higher order it _should_ be better.
For first order the errors are relatively small
for a normal stereo angle of 60 degrees. As order
goes up, the maximum stereo angle that will give
near to correct encoding gets smaller. Going above
that will just lead to the 'stereo over ambisonics' 
situation explained above.

For higher order, if you stereo signal is itself
the result of panning individual sources, you should
just skip the stereo step, and pan the same sources
directly into AMB.

Ciao,


# AmbDec configuration
# Written by MakeDec-0.7.0 at Tue Apr  7 00:04:23 2009

/description      5.0 First order decoder

/version          1

/dec/hor_order    1
/dec/ver_order    0
/dec/freq_bands   2
/dec/speakers     5
/dec/coeff_scale  fuma

/opt/input_scale  fuma
/opt/nfeff_comp   input
/opt/delay_comp   off
/opt/level_comp   off
/opt/xover_freq     300
/opt/xover_ratio    0.0

/speakers/{
#            id      dist     azim     elev     conn
#-----------------------------------------------------------------------
add_spkr    LS     1.500  110.000    0.000    alsa_pcm:playback_1
add_spkr    LF     1.500   30.000    0.000    alsa_pcm:playback_2
add_spkr    CE     1.500    0.000    0.000    alsa_pcm:playback_3
add_spkr    RF     1.500  -30.000    0.000    alsa_pcm:playback_4
add_spkr    RS     1.500 -110.000    0.000    alsa_pcm:playback_5
/}

/lfmatrix/{
order_gain     1.00000  1.00000
#            W        X        Y
#-----------------------------------
add_row    0.42898 -0.32754  0.26524
add_row    0.22140  0.24055  0.37042
add_row    0.08957  0.21362  0.00000
add_row    0.22140  0.24055 -0.37042
add_row    0.42898 -0.32754 -0.26524
/}

/hfmatrix/{
order_gain     1.00000  1.37000
#            W        X        Y
#-----------------------------------
add_row    0.67181 -0.19974  0.22776
add_row    0.40690  0.17925  0.23360
add_row    0.23133  0.29002  0.00000
add_row    0.40690  0.17925 -0.23360
add_row    0.67181 -0.19974 -0.22776
/}


/end


-- 
FA

Io lo dico sempre: l'Italia è troppo stretta e lunga.