[BC] AM audio bandwidth - AES Journal article

[Peter Smerdon]

There's an interesting paper published in the latest (May) issue of the
AES Journal on listener perceptions of AM audio bandwidth (in the
presence of interference). This looks to have come from NPR Labs, given 
the authors' backgrounds.

I'd be interested in any comments on the paper - or the subject generally.

My "takeaway" from this paper are:

1. AM radio audio bandwidth is still lousy - even with NRSC. Their 
"large sampling of receivers representative of participant sets" yields 
a mean -3dB point of 2450Hz and a mean -10dB point of 4100Hz.
Now that's "real mean"!
These figures are from tests which include NRSC pre-emphasis!
Unfortunately they don't disclose actual set quantities or brand/models.

2. The preferred bandwidth is 7kHz (not 5kHz ... hmmmm...)

3. The interfering signal used was USASI noise. How analogous (not 
digitus!) is that to real-world adjacent channel splatter in the North 
American environment.
(Note, I don't know about this one, so I'm asking my North American 
friends with more experience in this stuff.)

In deference to AES copyright, I've just included the paper Title and 
Abstract below.
For those AES members who haven't yet got their May issue, it's also 
available online at www.aes.org
I didn't see the paper available at www.nprlabs.org - but I think AES 
prohibits secondary publishing of Journal papers (at least for a while).

Here's the detail....

Participant Testing of AM Broadcast
Transmission Bandwidth and Audio
Performance Measurements of
Broadcast AM Receivers
by
ELLYN G. SHEFFIELD
Psychology Department, Towson University, Towson, MD 21252, USA
and
JOHN KEAN
NPR Labs, National Public Radio, Washington, DC 20001, USA

Abstract:
In late 2004 the NRSC’s AM Broadcast Subcommittee formed the AM Study
Task Group (AMSTG) to determine whether consumers would reliably
perceive audio quality differences of AM transmissions recorded through
commercially available receivers at bandwidths lower than 10 kHz. A
study was conducted to document the audio performance of a large number
of AM receivers and evaluate three prototypical receivers in a
consumer-based listening test.
Results suggest that for compressed audio that masks background noise
(that is, rock music), participants heard little difference between 7
and 10 kHz, regardless of adjacent channel interference conditions.
However, for less compressed audio that does not mask noise (that
is, speech) larger differences were heard, depending on interference
conditions. Results suggest that in general consumers preferred lower
transmission bandwidths to higher transmission bandwidths in real-world
noise conditions.

Regards,

Peter Smerdon
Melbourne, Australia.