Chronology Current Month Current Thread Current Date
[Year List] [Month List (current year)] [Date Index] [Thread Index] [Thread Prev] [Thread Next] [Date Prev] [Date Next]

Re: [Phys-l] Digital TV signal delay




John, you didn't use the word I used (buffering), and you gave a different
explanation about the need for buffering, but when you said...

They leave out most
of the detail on some frames, and interpolate. But they can't
do that until they've seen what comes after the affected frame.

Buffering is quite different from delays due to processing. Actually both
happen. Since the processing might be quick one must buffer or delay some
frames to match the delay caused by the slow processing. The basic delay is
probably due to the processing, but buffering must be used to make sure that
the frames are delivered at exactly the correct rate. The amount of
buffering needed to interpolate a single frame is very small, and is
probably imperceptible.

The same thing happens with HD radio. A regular FM station on one radio is
heard before the same material in HD. The processing for audio is much
simpler than for video so it sounds like a very long echo, and is most
disturbing. However the radio stations have around 8s delay (as I recall)
in transmission due to the heavy processing that must be done before the
signal goes out. So they delay the FM signal to match the digital signal.
This allows the receiver to switch between analog and digital as needed for
good reception. Apparently the HD radio introduces a slight delay in the FM
signal to match the internal decode time. Then they set all of the clocks in
the station about 8s fast so if the announcer gives the time, it will be OK
when received. So the time to encode is around 8s and the time to decode is
a fraction of a s.

Now that I have given the reason for a long delay in HD radio, I suspect
that a long delay in digital TV is similar. There is no necessity to delay
the analog video signal to match the digital video source because the TV
system does not switch between analog and digital to maintain quality. So
the analog signal will be received before the equivalent digital signal.
This "problem" will go away when the analog signals are shut off. However
the networks will have to time their broadcasts so as to account for the
delay in processing the outgoing signal. The announcers will have to look
at clocks on the wall, instead of their watches. If they have a computer
monitor, they may have to jigger the clock display on it.

The necessary processing to decode the compressed signal is significantly
less than the time to encode, so the time delay is most probably in the
encoding part of the chain. The encode time might be shortened by faster
processors, but then again it might not as they may need many seconds of
lookahead to achieve maximum compression with high quality. The delay in
the receivers while it is smaller may be standardized so that all receivers
are synchronized.

So the delays due to compression and decompression may never go away even
with superfast processors. It used to be that if you watched a concert on
TV or listened to it on radio you heard the music before the people in the
back of the hall, but with digital compression, this is no longer true.

I wonder if people will take advantage of this by having fast private analog
feeds of sporting events so they can place bets on sure things before the
digital signal arrives. This will not cause problems if all bets are placed
the necessary number of seconds before the event starts.

John M. Clement
Houston, TX