[LAD] Attenuation of sounds in 3D space

[JohnLM]

On 2010.07.29. 15:20, Gene Heskett wrote:
> On Thursday, July 29, 2010 08:52:04 am Jörn Nettingsmeier did opine:
>
>> john,
>>
>> On 07/29/2010 02:35 PM, JohnLM wrote:
>>> So from what I can understand, if I apply doppler effect, distance
>>> attenuation and other pre-process filters to a mono input and then
>>> push it through the panner plugin I get fairly correct representation
>>> of sound in 3D space. Right?
>>
>> depending on what you mean by "fairly correct", "representation", and
>> "3D space" :-D
>>
>> doppler would come into play only if you change the distance of the
>> sound, not when you move it on the sphere, as the distance remains
>> constant.
>
> This use of 'doppler' I would call incorrect, because the doppler shift is
> a shift in the apparent frequency of the sound rising at it approaches, and
> decreasing as it leaves.  I'm sure you have a term for what you mean, but
> doppler isn't it.  That police officers radar gun measures your speed by
> listening to the echo from your vehicle, and comparing it to the signal its
> sending, which to simplify, results in a beat frequency which is exactly
> your speed if approaching or departing exactly to or from the radar guns
> position.  That is why it is often called doppler radar&  the weather guys
> us it also.  Because they don't stand directly in front of you to take a
> reading, there is some small vector error in your favor.
>
> Someone else was trying to describe the distance vs square law change in
> the apparent volumes.  So let me try from the broadcast engineering field to
> explain that better.   Imagine a point src of energy, be it light, sound,
> or other radiation such as a radio or tv signal.  Measured at distance x,
> you will get your reference signal, call it 0 db in this case.  Now,
> without changing anything else move your measuring instrument to a point
> that is now at a distance of 2x.  You don't get half the signal, but 1/4 of
> it, because the same energy that was hitting a square of any arbitrary
> measurement, say a square inch, has in addition to being spread twice as
> wide at distance 2x, it is also twice as high.  So the new reading will be
> -6 db compared to the original '0' db.
>
> That is why we call it the square law.  The only way to get that back is to
> make the receptor itself 4 times bigger.  But while I have observed that
> there are quite wide variations in ears, I have not seen an individual with
> expandable ears (yet) :)
>
>> when you hand-craft distance cues, you should not expect wonders for
>> sounds originating inside your sphere of speakers. travelling through
>> the center quickly can be made to work, though.

I am fairly sure 'doppler' *is* what I meant. I work on spatialization 
project where sound sources can and will move around on arbitary 
trajectories in virtual space.

Well as for 'distance law'. Microphone receives and registers sound 
pressure instead of sound intensity.

intensity != pressure
intensity ~ pressure^2

While true that energy at 2x distance is 1/4, the pressure excerted is 
1/2 neverheless (pressure value is independent of area - i.e. it uses 
constant nominal area).
So you end up with a linear correlation instead of square one.

pressure ~ 1/distance
intensity ~ 1/distance^2

There is also the thing with acoustic impendance. For simplicity I 
didn't include it into "equations" and that's the reason they're not 
equalities, but are proportionalities.


>>> I failed to find anything discussing mixing in AMB. Can I just sum the
>>> channels of all sounds, like I would do it to any
>>> "direct-speaker-to-channel" formats?
>>
>> yes. that's a fundamental property of all linear systems, and
>> independent of the signal representation.
>>
>> i've written a little howto for ambi mixing in ardour a while ago, maybe
>> you'll find it useful:
>> http://cec.concordia.ca/econtact/11_3/nettingsmeier_ambisonics.html
>> for lac2010, i tried to find out how ambisonic mixing can be applied to
>> a pop production:
>> http://stackingdwarves.net/public_stuff/linux_audio/lac2010/Field%20Repo
>> rt-A_Pop_production_in_Ambisonics.pdf
>>
>> best,
>>
>> jörn