[LAU] ambisonics: fons ambdec and muse.demon.co.uk ambidec

Jörn Nettingsmeier nettings at folkwang-hochschule.de
Fri Jan 7 10:10:41 UTC 2011


On 01/07/2011 10:01 AM, Arnold Krille wrote:
> On Friday 07 January 2011 09:10:41 Giso Grimm wrote:
>> On 01/07/2011 12:33 AM, fons at kokkinizita.net wrote:
>>> Azim and elev are in radians, to convert to degrees multiply by
>>> 180/pi. Note that for all layouts on the demon page, dist will
>>> be 1 for all speakers. For ambdec you should use the real distance.
>>
>> in ambdec, is the distance used for anything else then delay and gain
>> compensation? Is it correct that for regular setups this does not matter?
>
> I imagine it could be used:
> Ambisonic when shipped seems to be normalized to distance of 1 unit.
> When your speaker array is much bigger or smaller, one could use that
> information to rescale the sound reproduction.
> But I haven't actually thought about this or even tried some formulas... Might
> not work.

i guess you are hinting at jerome daniel's thoughts on near-field 
compensation? i guess his point was that you have to assume a "standard" 
distance when encoding, and the decoding would be wrong if your diameter 
is different. however, as you say, since the decoder knows the assumed 
standard, it can correct accordingly.

for those non-ambiheads wishing to follow this discussion:
the near-field effect in ambisonics results in a bass boost. it's caused 
by the speakers. ambisonic theory (without NFC) assumes plane waves, 
i.e. speakers that are very far away, so that the wave fronts are not 
curved. since they are curved in practice, you get the bass boost. the 
same thing is responsible for the well-known proximity effect on 
directional microphones.

> Finally, one of the main advantages of ambisonics compared to that 5.1/7.1-
> crap is that your setup doesn't have to be perfect. Different distances and
> even different irregularities in shape can be compensated and still reproduce
> the signal as intended.

well, yes and no. the failure modes are different, too. depending on how 
5.1 and 7.1 are mixed, you can usually place the speakers any old way 
and still get something out of it, only the source positions will be 
displaced. heck, ever seen those setups with all 5 speakers below the 
screen, sitting on top of the vcr? not much surround, but you still get 
the idea of the mix.

in ambi, it tends to work well for minor displacements (which you can of 
course also correct in the decoder, but i'm assuming user errror here), 
but for large uncompensated placement errors, the whole reconstruction 
will fail. in this latter case, ambi will behave far less predictable 
than any discrete speaker technique.

and people should not expect wonders from ambi rigs. the ITU 5.1 setup 
is the most irregular you should try - anything worse than that, and 
ambisonics won't be much fun.
maybe when franz zotter and his friends from graz get their partial 
spheres decoder into a usable shape (right now it's pd magic only).
it basically enables you to build only parts of a loudspeaker sphere and 
discard all direct sound coming from the wrong directions.


More information about the Linux-audio-user mailing list