Subject: Multi-speaker ...
From: KEVIN AUSTIN (KAUSTIN@vax2.concordia.ca)
Date: Mon Mar 19 2001 - 07:38:26 EST
Sylvi (et al)
My focus on learning about psychoacoustics, and in particular ASA is
related to my feeling that there is aften some confusion about 'what
loudspeakers (can) do'. IME experience, multi-speaker and multi-channel
pieces need to work on their own terms, and these terms benefit from
being informed by a number of aspects of psychoacoustics. (In
non-akadameze -- it's the meat not the motion.)
Briefly (?) in ASA, Bregman proposes that the mind (attempts?) does two
'fundamental' things with the two incoming streams of information: (1) it
works to 'integrate' sounds (fusion); (2) it works to segregate and stream
(channelize) sounds.
[ WARNING: Extreme simplification ] A simple example from 'music' is
listening to (say) a bassoon and an oboe playing a duet. Let's say the
first note is a unison. The ear (mind's ear) works to sort out that there
are two -- and only two -- sounds present. Since the two players have
different rates of vibrato and micro-amplitude changes, the mind
determines (sic) that one set of partials is moving at the same rate, and
therefore 'integrates' these elements into 'one sound' -- and considers
it to be a coherent 'source'. It does the same with the other.
The two instruments pay a short rapid passage in unison. The
(well-trained?) ear hears 'one line' (stream) played by two instruments,
and can follow each one independently (even if played through a single
loudspeaker).
Change the oboe and bassoon for two identical pianos. Repeat the above.
While a well-trained ear wil be able to determine (possibly) that there
are two instruments (even if played 'exactly' together), it is probably
almost impossible to figure out 'which part' of the sound belongs to
'which' piano. While each piano note is 'fusing' with itself, since the
piano is effectiely 88 instruments sharing s common soundboard, it's
going to be very difficult to 'stream' the instruments with any degree of
security.
Let's move this to a soundscape environment.
You are in a city which has built two identical churches 100 meters
apart. They each have an identical set of bells. (Sorry, I need to
carillon with the over-simplification.) Both sets are ringing. At the
distance of 1 km, as you listen to these bells, can you work out which
bell belongs to which church?
While it may be possible to hear that there are a number of different
bells, it will be difficult to 'segregate' (stream) each set individualy.
Were the bells to be (say) two octaves apart from each other, a listener
would be able to hear the 'high bells', and the 'low bells', and stream
these into their own 'channels'.
As the frequencies of the bell sets get further apart, the range of the
frequencies (which were the same at the start) move far enough apart that
the ear hears two separate 'channels' -- they have exceeded the 'critical
bandwidth' required to be streamed. (The piano example is to play a two
part piece where the parts overlap -- very difficult to follow 'each
part' with confidence -- and then play the two parts separated by 2
octaves: beyond the critical bandwidth, the two parts can be heard (even
through one loudspeaker).
But how does this apply to multi-speaker projection / sound systems ....
In my experience, clarity of projection (or multi-speaker placement) is
strongly affected by the nature and quality of the source material(s) --
ie, the work is more strongly influenced by the sounds that by the
'delivery system'.
It may (in time) become 'second nature' to be able to determine whether
and 'how' a certain quality / type of sound will distribute / project. An
educated ear (whether genetic or trained) is probably required to tease
out the subtleties of this discipline. For those of us who are mere
mortals, not blessed with this as a genetic feature, I recommend study
<sigh! -- such an academic [yikes!]>, and a fine place to start is Albert
Bregman's Auditory Scene Analysis, MIT Press (1990) >> AND >> be sure to
get the excellent CD's that complement the book.
Enuf for one morning ... I'll leave a ponder about 'holographic sound
representation' for some other time.
Best
Kevin
kaustin@vax2.concordia.ca
This archive was generated by hypermail 2b27 : Wed Jun 11 2003 - 13:06:56 EDT