Fw: sursound: The pinna and Ambisonics/cross post

[Peter Lennox <peter@xxxxxxxxxxxxxxxxxxxxxxxx>]

 Hello,
My apologies for cross-posting; I would appreciate the wider perspective
that cross-posting might bring to bear on this problem: Most of the world's
audio engineers proceed on the assumptions mentioned below by Dave Malham,
namely that humans are able to more accurately locate a sound-source when
that source contains significant energy at below 1000Hz. The research
mentioned below is also used to support the argument that sound sources NOT
containing 'low frequencies' are correspondingly difficult to locate. From
personal experience, whilst I might well agree in the special case of sound
generated by artificial devices (speakers), such as electronic tills, mobile
phones, etc, such devices are ( from an evolutionary point of view)
comparitive latecomers, and can be considered 'minority' special case
audible objects. In the wider sense of the audible objects to be found in
real environments, it does seem to me personally that there is no sense in
which Lf facilitates location-identification better than Hf; - quite
possibly the reverse. This is a problem I've puzzled over, on-and-off, for
some years.
Is anyone aware of specific (and recent) research on this subject, or
theories which incorporate perception-as-related to 'real environments'
which speculate in this area?

thank you for your time
ppl
Peter Lennox
Hardwick House
tel: (0114) 2661509
e-mail: peter@lennox01.freeserve.co.uk
or:- ppl100@york.ac.uk

----- Original Message -----
From: "DG Malham" <dgm2@york.ac.uk>
To: <sursound@darkwing.uoregon.edu>
Sent: 23 October 2000 10:56
Subject: Re: sursound: The pinna and Ambisonics


Hmm - seems several people bit at the bait I dangled over the side. I had
noticed that several people were assumimng that the ear is better at
discrimating high frequency based positions so I baited a hook and waited
to see what would happen. My answer to those who think this is that they
should look at the research. As early as 1958, Mills, the originator of
the term "Minimum Audible Angle" (Mills, A.W. "On the Minimum Audible
Angle" JASA, VOL 30, NO. 4, PP 237-246) had shown that, the ear could
discriminate between two identical sound sources 1 degree apart for
frequencies up to around 750 - 1000 Hz. However, he also showed that with
significantly higher frequencies and at angles where the pinnae start to
have a major contribution to position sensing, the MAA increase to 30
degrees or more.  Later studies confirm this, for sounds without
significant features at lower frequencies where other hearing mechanisms
come in.

Mills used tone bursts, which were largely, but of course not wholely,
limited to the upper partials. Others who have used more continuous sounds
(sine waves, narrow band filtered noise) show similar or worse MAA's.
Where high frequency results have been given which are significantly
better than this (and sometimes better even than at low-mid frequencies up
to 750-1500 Hz) it soon becomes obvious that the signals used (pulse
trains, filtered music)  contain significant cues at frequencies where
other hearing mechanisms (ITD, IID) come into play. Deep intuition or not,
the facts simply don't bear out the argument.

What I would agree with, and what the research base DOES bear out, is that
if the high frequency, pinnae based cues are right, they can play a very
important role in enhancing the MAA resolution of the low frequency
ITD/IID mechanisms.  Conversely, if they are wrong (ie, in disagreement
with the low frequency cue) they will degrade the low frequency ones. If
they are simply confused, they will be rejected. This may well be related
to survivsl mechanisms whereby high frequency cues are far more likely to
be corrupted by nearby objects diffracting or reflecting them than low
frequency ones (the average tree trunk has little effect on low frequency
sound waves, which just flow round it, but can have pretty significant
effects on high frequency sound waves due to diffraction - and to a far
lesser extent, reflection - effects). When discussing any perceptual
research, a great deal of care needs to be taken to ensure that what it is
that is actually being tested is fully understood before making
assumptions. This is made doubly diificult because it is usually
impossible to reduce the perceptual "crosstalk" inherent in the geshtalt
that is the average human being's view of the world.

In the same vein, I like to point out that I feel there is as much of a
problem at the bass end and that many of the MAA measurements have to be
very suspect in this area since the common use of headphones, rather than
speakers (Mills used speakers and an anechoic chamber) mean that whole
body sound perception mechanisms (body cavity resonances, bon conduction
etc.), which are vitally important in the bass region, are simply not
brought into play.



On Sun, 22 Oct 2000, Peter Lennox wrote:

> Dave,
> I'm sorry, I've got to chip in here; there's an implication in your mail
> that, when push comes to shove, humans are terrible at locating sound
> sources which do not generate ......"low to mid frequencies". Although you
> don't specify the frequency ranges, I've  notion that (given your
> familiarity with Bamford et al), you mean up to 1700hz (ok, 2kHz max).
> I'm sorry, I simply don't buy the idea that resolution of spatial
information
> declines with increasing frequency; I don't just mean that it seems
> counter-intuitive, I mean that accepted proofs as to the effiicacy of the
> hearing system due to phase -locking ( / : ITDs) resulting in MAAs
(minimum
> audible angles) of the order of >1 degree wen compared to MAAs at hf,
> utilising IIDs (intensity cues), is simply not comparing like for like.
> The comparison relies on assumptions such as "static perceiver/ no head
> movement".
> My contention is that: given an ambulant organism with a temporally finite
> 'short term memory buffer', it may well be that 'hf'-borne information
> actually yields 'better' spatial resolution with increasing hf.
> I know I'm 'spitting into the wind' here, but I've a very deep intuition
> that, (up to a point), increased hf positively correlates with
> increase-of-spatial-information-uitilisable-by-humans.
> There's also a non-scientific argument here, that those creatures we know
to utilise
> sound to yield real-time detailed spatial information, use extreme hf,
> rather than extreme Lf. - I mean Bats, Owls, and Dolphins. All of these
> apparently achieve quite sophisticated environment-navigation
> (/representation) using sound, seemingly equivalent to our own abilities
> using ambient light.
> My own feeling is that there is a fundamental distinction arising here
> between 'special case / static percipient' examples of so-called
> 'perception', and 'real' perception which is evidenced *only* by
> intelligent, mobile organisms such as ourselves.
> I know I'm coming over all Gibsonian ("Ecological approach to [visual]
> Perception) here, but I do feel that there is a genuine question as to the
> general applicability of laboratory-based conclusions to specific 'real
> world' problems.
> Actually, I'd like to cross-post to another audio list to glean some wider
> perspectives; would you mind?
> cheers,
> peter
> ----- Original Message -----
> From: "DG Malham" <dgm2@york.ac.uk>
> To: "Rolv-karsten Rønningstad" <rolv-k-r@frisurf.no>
> Cc: "Angelo Farina" <farina@pcfarina.eng.unipr.it>;
> <sursound@darkwing.uoregon.edu>
> Sent: 21 October 2000 20:01
> Subject: Re: sursound: The pinna and Ambisonics
>
>
> On Sat, 21 Oct 2000, Rolv-karsten Rønningstad wrote:
>
> > Yes, I should possibly have been able to understand this myself. Anyway,
> > thank you very much for clarifying this with your answer. Can one then
> > deduct that more real sources, i.e. loudspeakers, improve high frequency
> > localization?
>
> Yes, to some extent, since the errors add arithmetically (because they
> change from speaker to speaker) and the correct cues add geometrically
> (because they are consistent from speaker to speaker) so you can roughly
> half the problems by quadrupling the number of speakers. As using this
> many speakers also means that you can thing about using higher order
> Ambisonics - and even going only to second order puts you much closer to
> the ears limits - it is, after all, only capable of discriminating a 30
> degree minimum audible angle at high frequencies as opposed to between 1
> and 3 at lower mid frequencies - you are beginning to get into the area
> where things work in perceptual terms almost as well as they can without
> going to hyper dense transducer arrays or other advanced technologies.
>
>         Dave
>
>
>
>
>
>