[BC] HD Radio -- Folks we have to get it right!

[Michael Bergman]

Bill,

> Was the study done in real moving automobiles in real traffic with the 
> random BER muting/blending we can expect in real life, especially 
> in places like SF where the same multipath will cause these problems at 
> greater levels of incidence than say, west Texas? Or was it in a lab 
> environment,  stationary, with no artifacts introduced to simulate 
> BER muting/blending?  
> Just curious ... since if these studies don't emulate/simulate the real 
> world they're just PR flak....

Your question is addressed to Bob, but I'd like to chime in.

Our experience with OFDM-based technologies (like IBOC and Sirius) is that
the alleged cleanup of multipath artifacts is very real.  In fact, I would
recommend against a test of FM analog vs. IBOC with induced multipath in a
simulated channel as a listening test because it is too favorable to IBOC.

IBOC (like other digital technologies) exhibits a cliff effect when the
signal fades, meaning digital audio blends to analog when conditions are
poor.  Incidences of muting/blending are not the dominant behavior of the
technology.  Within the coverage area, the performance generally stays
digital without BER-induced artifacts.  Hysteresis in the receiver prevents
frequent mutes and blends, so if the signal is weak and terrain limited, it
blends to analog and stays for a while.  

Regarding SF specifically, I tested KALW (1.9KW ERP) in 2003 and found that
their version of "Multipath Alley" had only a single mute.  We were a little
surprised that we muted there, since we (and the transmitter) were still in
the city, until we listened to the analog FM in the same location--very
noisy.

Regarding the NPR testing, the clean environment that NPR and Dr. Sheffield
set up was very good for listening, and the individual was able to
blind-compare two audio samples until they were ready to say "A is better
than B".  Then they did two more samples.  IIRC, the tester compared two
codecs 10 times for each pair of WAV samples.   

Generally, I see codecs tested in pure, clean listening environments like
this.  This helps prevent the very issues you mention from letting codec
artifacts hide in the "noise".  

My company uses this kind of listening test, lab BER performance testing,
and real-world field testing in combination.  Generally, a pure lab test is
good for studying codecs, but as you mention, not enough in itself.

Mike Bergman