[sdiy] ot: rotating speaker simulation or stupid approach

[Richard Wentk]

At 09:34 07/07/2003 -0600, Ian Fritz wrote:
>At 02:56 AM 7/7/2003, Richard Wentk wrote:
>
>>>I would look at this differently.  :-)
>>>
>>>Suppose you have two sticks to dip in the water and an observer in line 
>>>with them. Now do two experiments:
>>>
>>>1.)  Dip the first stick twice with a time separation dt.  The two 
>>>pulses will reach the observer
>>>spaced by a time interval dt.
>>>
>>>2.) Dip the first stick and then the second, again with a time 
>>>separation dt.  The two pulses will now reach the observer spaced by a 
>>>time interval that is different from dt -- larger or smaller depending 
>>>on which side side of the sticks the observer is located.
>>>
>>>This is a simple and obvious Doppler shift.
>>
>>No it's not. You're confusing an impulse with a continous sequence of 
>>wavefronts.
>
>Sorry, but you seem to be the one who is confused.  We are assuming a line 
>of speakers (or delays) whose spacing is less that wavelength of sound 
>(long wavelength limit).  Any wave can be decomposed into a series of 
>impulses.  This is elementary response theory (Green's functions, 
>etc.).  If successive impulses are sent to successive speakers, then a 
>Doppler shift will occur, as I showed in my simplified example of two 
>impulses from two sources.

Okay, but that's a special case, which I mentioned at the end of my email. 
You're talking about a situation which is completely different to the 
original one, where someone was saying that all you have to do is pan a 
sound along a series of speakers and you'll hear dopplering. For some 
idealised special case where very special conditions apply you will indeed 
hear Dopplering. But I've been talking about real world effects, not 
mathematical abstractions.

>>You can't Doppler shift an impulse because it has an infinite frequency 
>>response to start with.
>
>Nonsense!  You just change the effective wave velocity of each Fourier 
>component and you will get a Doppler shift.  If you need to look at it 
>that way.  For a more general view, you need to understand that any wave 
>can be written in the form y = f(x - ct), where c is the sound 
>velocity.  Changing c by motion of the source, either actual or effective, 
>will produce a Doppler shift.

I'd like to hear how a Doppler shifted impulse sounds different to a 
non-shifted one, for velocities that aren't affected by relativistic 
effects. ;-)

>>A more accurate representation would be to replace each single dip with a 
>>vibrator (no, not that sort... - actually thinking about it, they'd work 
>>as well as anything :-) ) producing a continuous stream of wavefronts by 
>>dipping up and down. Or indeed vibrating.
>
>When the wavelength becomes comparable to the speaker (or delay) 
>separation, then what you say is correct.  Several people have already 
>tried to explain this to you.  This is where aliasing (or equivalently, 
>interference) effects will come into play.

In that case we agree. But I still don't see how it's possible to build 
something that actually demonstrates this effect in a useful way given 
real-world speakers and a real-world frequency response range of 20Hz to 
20kHz.

That's what *I've* been trying to explain to people, many times over.

Richard