Oh boy, another thread to rant my views about improper use of dialnorm
Here's a bit of history for those who haven't followed this debacle.
In the analog days, it was important to keep the peak audio levels from exceeding a certain point for transmission as it would cause over-modulation which can create interference to other signals. This was complicated further in using FM modulation for audio (which was used by radio and analog TV sound) by pre-emphasis. TV and radio stations would commonly have a peak limiter in the path to prevent this from occurring and to avoid getting fined.
Even with a peak limiter installed, the loudness of the audio could vary. Dynamic range compression allows a signal to have a level amount of loudness without excessive peak levels. Everyone has probably noticed commercials sound loud as they often use this technique. Stations often used companders to even this out a bit.
When the digital TV standard was being developed, it was recognized that its increased dynamic range capability could cause further problems with the variance of audio levels in either direction. Dolby AC3 (also commonly known as Dolby Digital) was selected as the coding for ATSC in part as it could contain information (metadata) about the audio program. This included how many audio channels are being sent (usually 5.1 vs stereo), how they should be mixed down to stereo and how loud it is. The primary loudness information is called "dialnorm" which is short for Dialog Normalization. Dolby recognized that dialog is what is required to stay more constant, so the dialog is used as a reference for how loud the audio is. This value is meant to be set for each program element. Each show or show segment may have one value, while each commercial may have its own value. At the receiving end the Dolby decoder uses this information to adjust the volume. The volume is actually only lowered, so the louder program material is lowered to the softest one to even out the loudness differences. Commercials would probably have their volume lowered the most. Because this is on a per program element basis, the dynamic range of the program element is not altered.
Dolby also recognized that not everyone would desire to have the full dynamic range of a program. Loud sound effects or soft conversation could be an issue in some environments. The Dolby system also has built-in compressors in the decoders which can be used by viewers if desired. The Dolby encoded audio contains parameters for two levels of compression, one more drastic than the other. The moderate one is usually called "LINE" while the more drastic is called "RF" or "midnight mode". This way viewers had the choice of dynamic range rather than being a one size fits all approach.
One may logically ask “Why not just adjust the audio level itself to make it more uniform?” That would work to lower the louder elements, but raising the softer elements could push the loudest parts, such as sound effects, past the upper limit. By using a volume control at the receiving end that only lowers the volume (an attenuator) to even things out, the loudest elements are can be allowed to pass without limiting. There is also some politics about changing the levels on delivered material, and the dialnorm approach sort of gets around that by sending the audio itself unaltered.
It was a great idea, but it had one major requirement: The programs and commercials need the proper dialnorm value assigned to them. This would require extra support from the networks and stations. However, even if they did take all the time to do this, there wasn't a standard method of passing the information though the production and air playout process as Dolby encoding is not usually used for those steps.
What happened was that not only did the networks and stations not assign a dialnorm value to each element, they used a range of different static values! For years NBC used a low volume setting of -22db while CBS was at the opposite end of the spectrum using the highest volume setting of -31db. Not only was the variance between various programs not helped, but now there was a major difference when going from channel to the next (especially between CBS and NBC). Broadcasters took a system that was designed to minimize the volume differences and used it to make things even worse! Since then CBS has lowered their volume 4db to -27db and NBC raised theirs to -23db. 4db difference is much better than 9db. Most stations appear to used the default -27db value.
Up until recently the HD audience was the minority of viewers. However, as a result of HD growing in popularity and the analog shutoff, the HD transmission is now what nearly everyone is watching. Those getting SD from cable, satellite and converter boxes are likely getting downconversions from the HD transmission. Those volume differences which were annoying to just the HD viewers are now affecting everyone. As dialnorm is still not being supported properly despite Dolby's best effort to create products to aid in this endeavor, digital TV got a bad reputation for volume level variance. Not a big surprise and was predictable is that stations are now resorting dynamic range compression, and some in a bigger way than ever before.
Now this may sound like the stations are the villains, and indeed some have gone a bit overboard, but they are also the victims. The broadcast industry did not make dialnorm matching a priority. There still isn't a widely used standard that passes dialnorm through the production and distribution process. Stations get material that is all over the map for loudness including network feeds. The FCC requires that the dialnorm matches the program level, and compression is a practical and quick fix. I could see NBC stations with their lower volume dialnorm trying to compete with other louder stations (due to louder dialnorm) by adding more compression. As I've mention in the past, CBS and NBC stations can bypass the network metadata dialnorm on their DP569 encoders, but since CBS has lowered their dialnorm volume it's only the NBC stations that really need to consider this.
The bottom line is that adding compression will get less complaints from viewers than excessive dynamic range and wide loudness variations between program elements. The FCC only mandates that the dialnorm matches program content, not that it sounds good. As the OP noted, we may have taken a big step backwards from the analog days in this respect.
Does anybody care? Unfortunately I think too few do, and most won't miss what they don't know is missing. I mentioned to one well known manufacturer at a trade show that their audio program leveling product altered the original intent of the creators, to which they said “So?” It's sad but not surprising, and I expect it to get worse before it gets any better. The good news is that a system is in place to fix it, and complying with its intent will give back the full dynamic range that was intended with digital audio transmission. The bad news is I doubt there is much incentive to do so, especially in a down economy. Coupled with robbing HD video bandwidth for subchannels, it's not good news at all for much of OTA ATSC technical quality.
I guess FOX's splicer technique is one guarantee that stations can't compromise the network feed, at least if it's used as it supposed to be without any downstream re-encoding or rate shaping. It will be interesting to see how long CBS will continue to restrict their O&Os from multicasting. It's sad to read that one of their major O&Os is using audio compression, and even worse that it's a duopoly.
Sorry about the length of the post, but it's one of my major pet peeves too.