Revenge of the Couch Potato

Our Story To Date. Alert readers may recall my recent concern with the variance in audio levels as experienced by us home viewers. In my last piece on this issue ("Metadata in the Home: A Case History," March 10), I discussed my own personal "couch potato" history.

If you recall, I told you I couldn't make the system work perfectly (the set-top box from Scientific Atlanta and my Bang & Olufsen Avant TV did not successfully communicate regarding the presence or absence of digital audio, requiring me to restart the TV any time I wished to enjoy digital, as opposed to analog, sound). I also noted that there was essentially no information about the Dolby metadata system available from either the set-top box manufacturer or the cable service provider, in this case Charter Communications.

These things, in and of themselves, are not an overwhelmingly big deal. They are annoying, and they make it fundamentally impossible for me, as the end-user, to optimize my system.

In all fairness, the net effect is not all that bad - the analog sound isn't significantly worse than the digital sound, particularly in the face of the overall broadcast audio quality (which is okay but not great), and the lack of information about metadata reduces the potential for the system to some degree, but does not render it totally useless. I have confidence that the technology will settle out eventually and the long-promised capabilities will be there.


Through the good offices of Jeffrey Riedmiller of Dolby Labs, I was able to borrow a Dolby LM100 level meter to measure the actual dialnorm metadata values and LeqA levels of my particular installation, and have now collected a fair amount of additional information about the whole system. I spent several days channel surfing and getting a real feel for how the TV audio system works in terms of levels, using the same measurement instrument recommended for all content producers and service providers.

I haven't yet begun to measure the differences in level between commercials and program material, so I won't go there today. However, I took a lot of measurements of all the channels, analog and digital, using both RF mode and line mode, and also observing the direct RF cable feed.


This all follows some informal measurements I made about a year ago, where I simply took quick relative average level readings from all channels, and was appalled to discover that the range of observed levels was about 18 dB; the standard deviation was about 5 dB (this suggests that approximately one-third of the levels varied more than 5 dB from the average). These measurements correlate well with studies by Jeffrey Riedmiller and others at Dolby, as well as work by Michael Guthrie at Harmonic Inc.

Clearly, there was a problem a year ago. Sadly, there still is.

You all recall dialnorm, right? It's a metadata calibration value and is supposed to be set at the A-weighted power-based average of dialog level (aka "dialogue LeqA"). The default setting that Dolby ships it with is -27 dBFS, and the level at which no attenuation occurs is -31 dBFS, the lowest level it can be set. The content producer is supposed to set it at the same level as the measured dialog level.

On the digital channels with dialogue where dialnorm is present, I measured dialog levels ranging from -40 dBFS LeqA to -17 dBFS LeqA. Unfortunately, I measured only two dialnorm settings: -27 dBFS (the default, encountered on 75 percent of the channels measured) and -31 dBFS, the setting that defeats any static attenuation via metadata.

This is bad news indeed. Only three out of 28 measured channels were correctly set within a dB, and only 10 made it within +3 dB! Further, I suspect these were mostly due to dumb luck. Meanwhile, seven channels were more than 10 dB off! Seventy-five percent of the content producers aren't bothering to even set metadata values, and those that are setting it are setting it to a level where it is supposed to do nothing, which is to say they are in essence attempting to defeat the system! In fact, the system is defeating them, by miscalibrating their levels by the amount of the error.

As our visually-oriented friends might say-it ain't a pretty picture!


It's not all bad, though. On the analog side of things, levels from my provider, Charter Communications, have gotten significantly more consistent. A year ago the range of levels of the measured analog channels was 14 dB, and the standard deviation was about 3 dB. Today those level ranges have been reduced significantly, for a range of only 7 dB and a standard deviation of only 2 dB-much, much better! Not only that, but it is also good enough to be really acceptable! There's not much need for further improvement in this realm.

So, there's hope. A little attention to levels can actually make a significant improvement.


The system still isn't working, though. For it to work, everybody has to play and play by the rules. Right now, the content providers are not yet using dialnorm, which is to say that they are mis-using it.

Based on my experience and the letters I've received, the service providers and their vendors are not telling their end-users what they are doing or what the end-users should do to optimize their systems. Further, all end-user systems don't work together as the seamless transparent media delivery package that Dolby rightly desires and envisions.

Efforts to contact my service provider were unsuccessful (Charter Communications simply never picked up their phone when I called, leaving me to listen to "exciting" announcements for 25 minutes). At Scientific-Atlanta, I was able to speak to Jim Kiker and S-A's audio expert Dave Sedacca. They had enough to say about the problem that I am going to have to take up Scientific-Atlanta's side of the story in another column.

They also tactfully suggested that I continue to pursue Charter through its national headquarters, so stay tuned on that front. In any case, they were very clear on the issues and have been working hard to (a) help their customers (cable operators) deliver better service, and (b) improve the quality and performance features of their set top boxes.

Finally, Bang & Olufsen noted that the company has encountered my digital-vs.-analog signal problem with some other set-top boxes and service providers and is working on a fix, taking the position that it is B&O's responsibility to make the TV talk to the set-top box because the company knows full well that the set-top box manufacturers are not going to adapt to Bang & Olufsen!

What we can take from this is that the people who supply the media to us may not necessarily be good at nor devoted to good end-user relations or communications, but they really do seem to know what they are doing, with considerable depth, thoughtful and technical expertise.


Dolby needs to exert some influence here (in all fairness, the company already is working hard at it behind the scenes). Dolby needs to insist (a) that the content providers use the system correctly and (b) that the media vendors accurately and fully share with their end-user customers and each other what is going on. The company need to emphasize that the various players really need to actually do their parts in this quite elaborate production process.

The sooner the better, I hope.

Thanks for listening!

Dave Moulton