|
There are several issues here:
1) different mics are known to behave quite differently in
'reactive' (ie enclosed or semi-enclosed)spaces (ref: Angelo Farina), and thus
vary in their abilities to reproduce the salient plane-wave characteristics.
Allthough the same mics are being used to record human speakers and
loudspeakers, which would seem to rule out this factor, because of the ways that
the two types of source may differ in their relationships with local surfaces,
this might still be pertinent. (in other words, you could get some mics which
yield similar results for both types of source, theoretically).
2) Loudspeakers sound more like loudspeakers than they do
anything else. A given speaker's diffusion characteristic, and hence it's
acoustic relationships with local features remain fairly constant, irrespective
of programme material (to a point!).
3) A loudspeaker is (generally) designed to be part of a
'sound field reproduction system', rather than as a sounding object in itself
(with the exception of speakers which are in fact designed to be part of a
musical instrument, such as guitar cabs - these could be called 'production
systems' rather than re-production systems). As such, a design
intention is to minimise resonances at all frequencies, whereas all the
'sounding objects' the speaker is to imitate rely on resonance for the
majority of their energetic output. Because of acoustic coupling
(air-transformer) effects, an object in resonance interacts with it's local
environment quite differently from a sounding object which is not in
resonance.
4)In fact, from 3), an important design criterium for speaker
enclosure design is to minimise speaker -room interactions, inhibiting speaker
localiseability in favour of phantom image localiseability.
It's my guess that the loudspeakers should be most
localiseable when they are a) distorting in one or more ways
(compressing,clipping, resonating etc.,) or b) playing material which is
electronically generated, and never in itself contained any 'spatial
information' (with apologies for using the term 'information' in this
way).
This highlights the problem with the (generally held) notion
that it is possible to accurately record 'a sound' yet strip it of any spatial
attribute - what is often called 'direct sound'.Consideration of what might
constitute 'direct sound' shows the term to be a theoretical entity which is
useful, but about as likely as a 'point source'.
It is unclear whether the sense of moving through an
environment that you are after would be improved by better localisability of
sounding objects in that environment, or whether some other
characteristic of the environment itself might directly appeal to spatial
perception in such a way as to generate a sense of 'out there-ness'. (I suspect
the latter).
Lastly, from an ecological point of view, there's the question
of whether a highly symmetrical environment such as the corridor you describe is
particularly 'natural'; did we evolve best acuity for this type of environment?
I suspect not, and that a highly symmetrical environment is second only to an
anechoic one in terms of difficulties in localisation. Reflections and
resonances are simply too homogenous, which is tantamount to saying 'lacking in
potential-information' .(I've done some informal experimenting in large circular
empty spaces)
The bottom line is, loudspeakers don't sound like 'real
objects' (unless they are actually being real objects). They actually
imitate a wide variety of real objects fairly well from the perspectives of
considering frequency response, dynamic range etc., but none of their design
criteria have anything to do with spatial perception.
It would be interesting to speculate as to what conditions you
would need to achieve similar results for 'real' speakers and
loudspeakers.
regards,
ppl
|