From the very beginning, the U.S. broadcast digital-television system offered some amazing capabilities. HDTV could be broadcast in six-channel (5.1) surround sound in viewer-selectable languages and with picture description for the visually impaired. Viewers could also select from different commentary channels broadcast with the same show. Unfortunately, there haven't been any such transmissions yet, and the reason bodes ill for the use of advanced compression in U.S. digital terrestrial television (DTT) broadcasting.
At issue is the installed base of DTT receivers, whether in digital TV sets or in stand-alone boxes. They are required to have MPEG-2 video decoders and AC-3 audio decoders.
The AC-3 circuitry can decode anything from a single monophonic channel up to six-channel surround sound, and other circuitry can direct the appropriate data stream to it for decoding. Thus, someone desiring English could select the English data stream, someone desiring French the French stream, someone desiring Spanish the Spanish, etc. Someone desiring English with English picture description could select that, too.
Unfortunately, even after compression coding, 5.1-channel surround sound takes up around 0.4 million bits per second (Mbps). That's not much compared to perhaps 18 Mbps for an HDTV picture, but the multiple audio packages add up. English, French, and Spanish, in picture-described and non-described versions, would add up to somewhere in the vicinity of 2.5 Mbps. Add 18 Mbps for HDTV pictures, and the result is 20.5 Mbps--too much for a U.S. DTT channel.
So the DTT standard allows for a single, language-independent music-&-effects (ME) surround sound package to be combined with dialogue (D) channels or channels for the visually impaired (VI), in multiple languages, instead of a complete mix (CM). Allowing the same 0.4 Mbps for the surround sound ME package, the six dialogue or description channels combined would add only another 0.4 Mbps, well within the channel capacity.
Decoding the D channel separately from the ME package requires dual-stream decoding, a relatively insignificant addition to other DTT circuitry, but still an addition. Manufacturers have chosen not to provide the second-stream decoder. As a result, multilingual DTT broadcasts in the U.S. are all but nonexistent.
Suppose that tomorrow all manufacturers have a sudden change of heart and install dual-stream audio decoders. Suppose they appear in stores in time for the 2007 holiday shopping season. When would broadcasters then be able to transmit ME+D audio instead of CM?
The answer could be never or at least a point so far in the future that it can't be forecast today. The problem is the millions of DTT receivers that have already been sold and the millions more that will be sold between now and the 2007 holiday shopping season.
Those existing receivers can decode only one channel. If a broadcaster transmits ME+D, viewers with existing receivers might be able to choose between dialogue and music-&-effects, but they couldn't get both. What broadcaster would be willing to lose audience that way?
At some point, perhaps 90% of the audience could have dual-stream decoders (again, that's assuming some great change of heart by manufacturers). Would a broadcaster be willing to give up on the remaining 10%? And how long might that 90% take? According to the Consumer Electronics Association, as of January of this year, 76% of American households had TV with stereo sound, almost 22 years after the first stereo TV sets went on sale.
Now consider the video decoder. Advanced encoders, such as MPEG-4 AVC or VC-1, should be able to deliver HDTV in many fewer bits per second than MPEG-2. But existing DTT receivers have only MPEG-2 video decoders, not MPEG-4 AVC or VC-1.
USDTV, which knew every one of its customers, decided to take advantage of the increased coding efficiency of MPEG-4 AVC by delivering pay-TV movies that way. The company's newer receivers had MPEG-4 AVC decoding built in. For the older receivers, they came up with a USB 1.1 MPEG-4 AVC adaptor that could easily be plugged in by customers, and they offered it for free to all of them.
That way they could be assured that all of their subscribers would have access to the content using the advanced coding. It was a plan that broadcasters, who don't know all of their viewers, will not be able to duplicate.
Shortly after showing their adaptor this spring, USDTV filed for bankruptcy protection.
Mark Schubin is an engineering consultant with a diverse range of clients, from the Metropolitan Opera to Sesame Workshop.
