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

[Robert Orban]

At 05:06 PM 10/15/2005, you wrote:
>From: "Bill Sepmeier" <dcpowerandlight at hotmail.com>
>Subject: Re: [BC] HD Radio -- Folks we have to get it right!
>To: broadcast at radiolists.net
>Message-ID: <BAY107-F16DD562427C022938F39ACA27C0 at phx.gbl>
>Content-Type: text/plain; format=flowed
>
>Bob,
>
>WHY is it that it's this late in the monopoly game before some of this
>industry's reputable players have realized how substandard IBUZ really is,
>as presently implemented? Is it because innovators like yourself have
>accepted a future that is less than today's reality, in a quest to sell
>add-on algorithyms that might silk-purse this sow's ear?  Is the need to
>sell more processors and widgets so great that you'll compromise real
>quality for net-present-value-zero quarterly accounting - even if the
>"improvement" sinks the "ship" that has been our industry as consumers
>reject still substantial bit error drop out, limited bandwidth and
>telco-grade audio on secondary channels in favor of XM, Sirius and superior
>implementations of digital radio?

Perhaps I am prejudiced against analog FM from living in the Bay Area, AKA 
"multipath central," for all of my adult life. I can't enjoy listening to 
our local class-B FM classical station (KDFC) in my car (which has a 
spatial diversity radio with two antennas) because of the incessant 
spitting and picket-fencing interposed with blasts of multipath distortion 
and a stereo image that is constantly widening and narrowing as the radio 
tries to make the best of a bad situation. Compared to that, the artifacts 
of 96 kbps HDC are nothing. No, it's not quite "CD quality." So what?

In the course of developing processing, I have spent a _lot_  of time doing 
direct A/B comparisons between the analog-FM and digitally processed 
channels in the audio processor. Without the codec, the digital processing 
channel simply blows the analog away in terms of low distortion, crispness, 
and definition of high frequencies. The codec does not preserve this 
perfectly, but it preserves enough to make a significant improvement 
compared to the analog-processed signal through an ideal transmission path 
having no multipath. 75us de-emphasis is down 17 dB at 15 kHz; this 
severely limits the HF headroom of the basic FM transmission channel in the 
US and it always will.

As far as XM and Sirius go, both of them use far lower bit rates on average 
than 96 kbps; neither sounds as good HD FM as 96 kbps. However, their 
limited bit rates do not seem to have stunted their growth.


>IBUZ, as Rich has aptly named it, was a political development, never a
>"superior technology," and is not a demonstrable improvement in quality or
>service over present or past analog; not based on what I've listened to
>anyway.... not with its "CD Quality" (LOL!!) audio streams running at
>less-than ISDN rates at best and with the continuing problems encountered
>with moving platform decoding ... it offers nothing new to the consumer
>except large additional costs and little to anyone else - except the
>licensers and manufacturers of IBUZ components.  Frankly, I'm a bit
>surprised that you, of all people in this business, a guy who made FM and AM
>sound as good as uncompressed digital years ago, has "adopted" such a
>science experiment so readily.

Unless average modulation were reduced about 10 dB compared to today's 
practice, FM cannot sound as good as uncompressed digital because of the HF 
headroom limitations. It doesn't matter whose processor is in use; it's a 
fundamental compromise made in 1948 based on the power spectral content of 
1948 program material. Today, it's a hopelessly obsolete design parameter 
that is unfortunately built into the transmission system specification.

As for analog AM, the average radio was down 3 dB at around 2 kHz about 25 
years ago, and measurements that the NRSC's AM working group has recently 
made indicates that not much has changed in 25 years. Without audio 
processing to do some pre-compensation, the average AM radio has poorer HF 
response than a toll-quality telephone circuit (except that the radio 
doesn't drop off as quickly above 4 kHz). The best audio processing in the 
world cannot push anything that would qualify as "high fidelity" through 
the average AM radio (which appears to be down around 15 dB at 5 kHz these 
days), let alone quality that could be seriously compared to uncompressed 
digital. The best we can hope for is "entertainment quality" that is 
adequate to the requirements of the program material.

>   If your processors sounded as marginal  over
>the years as IBUZ does today you'd not be as well known or wealthy ... so
>why hang your hat now on this almost dreadful turn of events called IBUZ?
>
>IMHO, maybe if Rich repeats himself enough, people will wake up and see that
>we're being "had" by a bunch of Reno gamblers and insiders - political folks
>interested in the monopoly money they might make instead of a viable
>industry that serves the public and it's own shareholders with quality
>broadcast services, not this "widget du jour" called IBUZ.

The "widget du jour" was, in fact, in development for well over 10 years 
and is a response to the NAB's rejection of Eureka 147 in the early '90s. 
Yes, it was driven by politics, as is virtually every other aspect of 
public policy in a republic. Given the political stipulation that the 
digital signal has to coexist with all existing analog transmissions in the 
same RF spectrum, one can't do very much better than HD Radio's modulation 
scheme, which is not state-of-the-art anymore, but is close; making it 
state of the art would only allow the digital-channel RF power to be 
reduced by perhaps 2-3 dB, which would not resolve the complaints that 
people have with the system regarding interference. Those are intrinsic to 
the requirement that the system be IBOC. If anything is "substandard," it 
is the concept of IBOC itself, not iBiquity's implementation.

The alternative is either the status quo or freeing up new spectrum for 
digital broadcasting.

Bob Orban