[BC] Balancing the processing from analog to digital

[Robert Orban]

At 07:27 AM 12/21/2006, Rich Wood wrote:
>------ At 02:23 AM 12/21/2006, Robert Orban wrote: -------
>
>>You missed my point. At least in mobile reception, there will be 
>>areas where a significant amount of crossfading occurs between the 
>>analog FM and HD1 digital streams. If the volume drops 10 dB each 
>>time the radio crossfades from analog to digital, this will be a 
>>HUGE irritant -- far worse than excessive compression or peak 
>>limiting. Indeed, I believe that this would cause 95%+ of the 
>>audience to tune out after it happened a few times.
>
>That's exactly my point. How do we deal with it? To sell a receiver 
>to anyone other than a radio station employee we need to show 
>there's a reason to buy one. I assume the Sangean is capable of 
>entertainment quality in digital. It doesn't move, so the 
>crossfading happens less often. My Accurian with an attic antenna 
>crossfades often, mostly with changes in the weather. The audio 
>section isn't good enough to notice. The NPR station here processes 
>gently, so I should notice the alleged advantage of digital. With 
>the Accurian there's no difference. I can force it to analog by 
>using its supplied antenna. When I connect the attic system it 
>sounds exactly the same. All that happens is the logo lights up.
>
>We have two incompatible scenarios here. Entertainment quality for 
>home systems and lack of irritation in a mobile environment.

Achieving loudness parity between the analog and digital channels 
still allows processing for the digital channel to sound very good 
even on a high resolution home system. Not only is there more dynamic 
punch than in the analog channel, but there is no high-frequency 
squashing and no clipper grunge. In my opinion, the element limiting 
the quality on a 48 kbps channel is actually the HDC codec and not 
the peak limiting necessary before the codec. I've done a *lot* of 
listening directly to the output of processing designed for digital 
channels in the course of designing and tuning equipment like the 
6200, 6300, 8500HD, and Optimod PC. It can sound really good while 
producing peak-to-average ratios that provide an appropriate loudness 
balance between the analog and digital channels on an HD Radio receiver.

Also, please bear in mind that on HD1 channels, the consumer hears 
the blend every time she changes stations. The reception always 
starts out analog until the HD buffer fills, at which point the radio 
blends to digital. This is true regardless of whether the reception 
is mobile or at a fixed location.


>>These days, who has the time or budget to carefully adjust the 
>>loudness level of every element (including each commercial) before 
>>it get entered into a playout system? In theory, this could be done 
>>automatically by analyzing each file's subjective loudness level 
>>and adjusting the file's level appropriately (which is NOT the same 
>>as peak-normalizing the file; peak normalization has nothing to do 
>>with subjective loudness). But in practice, this isn't what happens 
>>at most stations.
>
>I have to disagree. In the olden days things were often done on the 
>fly. In AM, in particular, riding gain meant pinning meter needles 
>and letting the processing take care of everything. Today, a cut is 
>loaded into an audio system once. There certainly should be time to 
>make sure the quality of that cut is the best possible. I find it 
>very hard to accept "good enough" when it comes to our only product.

I agree that this is the ideal. But I don't think it's happening in 
practice. Consider that some of the major television networks *still* 
have huge loudness disparity problems between various elements in 
their HDTV feeds to affiliates, despite the fact that tools have been 
available (from Dolby) to set the Dialnorm metadata for each element 
to ensure source-to-source consistency. We're talking about 
billion-dollar businesses who don't get this right. That's why I have 
my doubts about local stations, where staff has often been cut to the 
bone in the interests of efficiency.

Meanwhile, there is not a shred of evidence that tasteful processing 
causes loss of audience share. In my experience, insufficient 
processing usually causes more problems than too much processing. (It 
has been said that the BBC started gently processing their classical 
music service when one of the BBC directors complained that he 
couldn't hear the quiet passages in his Rolls :-)

>  We don't have time because most of the staff has probably been 
> fired. The other problem is the loudness wars going on in CD 
> mastering. How do you correct something that has absolutely no 
> dynamic range? How do you process it?

Minimally. But even hypercompressed CDs can benefit from low-density 
multiband processing whose main purpose is to automatically 
re-equalize their frequency balance to ensure a smooth, consistent 
flow between program elements in a broadcast.

Bob Orban