The following are some thoughts about Digital video and audio transmission in general, and Terrestrial Digital Television broadcasting, in general.  Yes, but I'll organize it better some other day...

First, it is NOT necessarily High Definition TV (HDTV.)  DTV (also referred to as Advanced TV or ATV) can include what is called HDTV, but, just like the graphics on a computer, there are many resolutions and modes possible.  Initially, you can expect everything to look pretty much as it does now...  If you are lucky...

Firstly, let's get rid of the wide-ranging misconception that "everything digital is better."  This probably stems from the advent of the compact disk:  To be sure, it sounds a lot better than a vinyl disk did (I'll concede that it is possible, with an extremely good recording and very good playback equipment, a vinyl disk can, in certain ways, outperform a CD...) by virtue of the fact that the media and the means of encoding the sound is more robust.  There is built-in error correction, and a reasonably wide dynamic range available.  I really do like CDs and I'm still amazed at the technology...

Some people argue that CDs were introduced too early.  They have only 16 bits (yes, under the right circumstances the shortcomings of this resolution can be demonstrated...) and the sampling rate is a bit low (yes, some people can hear above 22.05 KHz...) but overall it is very good.  There is, however, no attempt at compression.  A relatively simple lossless compression scheme could have resulted in either smaller disks, or ones that held several times as much audio as they currently do.  At the time of introduction, however, such a burden would have impeded their launching by adding extra costs.  Personally, I don't have a problem with a disk that size that holds only 74 minutes of music.

If CDs were being developed nowadays, they would be done very differently:  They'd probably be physically smaller and/or hold more information, they would likely have more than 16 bits of equivalent resolution, their sampling rate would be something higher, and they would implement some form of data compression.

Would they sound as good?

Well...  How good something sounds is often a basis for a endless, unproductive debate based on subjective quantizations and personal preferences, so I'll skirt that for the moment...

What about digital video?

Video is a different matter.  For audio, one could live with having a couple megabits-per-seconds streaming off a disk and getting reasonable results (i.e. the 650 megabytes or so for 74 minutes on a CD is quite reasonable...)  Video, on the other hand, if stored in a digital format that was largely uncompressed (we'll get rid of things like sync and burst, since we already know what they look like and when they will happen...) we'd get only 10-20 seconds on a CD.  Sure, it would look great, but you'd better not blink!

Enter video compression.  If you were to take your VCR and go through frame-by-frame, you'd notice that almost all frames look very much like the several frames before and after them (obvious exceptions being cuts, titles, etc.)  So, if you just sent what has changed between frames, you can drastically reduce the amount of what you are sending.  You can also save some space by identifying what portions of the picture have moved and say "Ok, this part of the picture is moving left at a certain speed" and you won't even have to send that portion of the picture, even though it is changing.  You can also say something like "Well, the human eye can't see that the colored blob over there on that part of the picture has shifted a bit, so we'll ignore that for now and worry about it when we think the eye can start noticing the change..."

All of these applied properly can result in very drastic reductions in the amount of data that you need to send to keep a picture updated.  These are also called lossy compression schemes since you aren't really sending the pictures, but elements of the pictures that, by the estimation of the compression algorithm, won't be noticed by the viewer.

These losses do produce visible artifacts, but how visible these artifacts are depends upon several things:

• What is your data rate.  The more data you can squirt out, the more completely you can keep track of the changes going on in the video, and the less difference there will be between your original picture and the one you are sending.
• The quality if your encoder.  The theoretically perfect encoder will be able to communicate exactly how the picture has changed to the decoder on the other end (given a sufficient data rate, of course.)  Good encoders can do "more" with "less" - that is, convey the changes in an efficient manner (e.g. with fewer bits.)
• Does the viewer know what to look for?  The average viewer is not even aware of many of the digital artifacts present on compressed video (most notably DVDs.)  If you do point them out to someone (and they are able to see them) then they are probably "ruined" for life (i.e. now that they know what to look for, it will probably bother them in the future...)

Now, in the case of audio encoding, we have to do things a bit differently.  Fortunately (for the video people, at least) the human eye is pretty slow and you can easily fool it into seeing what it you want it to see.  For example, most people can't see the flicker of a television picture, a movie, or fluorescent lights.  They eye can't really detect color detail on picture elements that are very small.  Our brain does a remarkable job of "filling in the gaps" between what our eye sees and what we expect to see (consider. for example, the big hole in our field of vision where our optic nerve is- the so-called "blind spot") and we should be very grateful indeed!  Our ear, on the other hand, is not nearly as forgiving.

It is possible, for example, to hear the one singer in a choir that is flat, a very slight "wow" of an off center record, or even the single female voice in a crowd of men.  What the ear is not very good at is picking out one particular sound among ones that are quite similar.  Similar to the eye, the ear also can't always tell how fast a sound changes (it depends on the type of the sound, of course...)

So with audio compression (like MPEG audio compression) what is done is to make a spectral "map" of the sounds being made, and then transmit the points on that map to the receiving end which, based on the available information, reproduces that map.  Like video compression, it also relies on changes over time to be somewhat slow (a note in a piece of music will last for at least a few tens of milliseconds, usually...)  Oh, since you might already know that MPEG stands for Motion (or Moving) Picture Experts Group, you might be wondering what pictures have to do with voice?  As I recall, motion pictures have had sound since at least the 1920's...  So there!

With sight and sound reduced to bits (and fewer of them!) you are ready to transmit them.  How much should be transmitted?  Too slow, and the sound and pictures have obvious artifacts.  Too many, and we are wasting bandwidth (and money!)

Having worked around MPEG (and similar) audio and video schemes, here is my estimation of how things look and sound at various bit rates (at NTSC resolution - for High Definition TV, these rates will not apply!)  These are rough ballpark figures, using "good" encoders:

For video:

• 1.5 megabits-per-second.  This, for the average person, may be on par with a VHS recording.  Since the nature of the degradation in video quality of a VHS tape is vastly different from those artifacts in a video compression stream, it is difficult to make a direct comparison and I am loathe to do so.  There are obvious artifacts if you know what you are looking for.  You might use this for news feeds, but someone will likely complain...
• 2 megabits-per-second.  This represents a surprising amount reduction in the artifacts as compared to 1.5 mbps.  I still wouldn't call it broadcast quality, though.  This is plenty good for news gathering, etc.
• 3 megabits-per-second.  This isn't too bad for most broadcast use.  With a good encoder, artifacts are difficult to see in real-time.  The only place where you are likely to see artifacts is in scenes with intense action and detail (i.e. sporting events, action scenes in movies, a wildlife scene where you have lots of shimmering water and waving grass, etc.)
• 4-6 megabits-per-second.  This is getting to the point of diminishing returns.  You can likely handle the shimmering water and action scenes with this most of the time.  The artifacts, when they do appear, will be visible for such a brief time that most people will never even notice them in real-time.

• 1.8-4 kbits-per-second.  Intelligable speech is certainly possible, but you may have a bit of trouble telling if the person on the other end is your girlfriend or grandpa at first using the lower data rates.  The (now defunct) Iridium phones operated in the 3-4 kbits-per-second range.   (Monophonic)
• 8 kbits-per-second.  This sounds pretty good, but has a "gravelly" or "watery" sound to it.  PCS-type cellular telephones operate at this.  Music sounds really terrible!  (Monophonic)
• 13 kbits-per-second.  This sounds similar to how the 8 kbit-per-second sounds, but has markedly less "graveliness."  This is the rate at which digital "cellular" phones (not PCS) operate.
• 16 kbits-per-second.  Great for speech (only a little bit of "wateriness") - OK even for voice interviews on broadcast.  Music sounds better, but still sounds "watery," and the frequency response is on-par with your average AM radio.  (Monophonic)
• 24-30 kbits-per-second.  This is sounding good enough where many people can't tell the difference between  it and an equivalent analog line.  You could get away with using this as a backup audio link for an AM radio station and most people wouldn't notice, but music still has some "rough" edges on it. (Monophonic)
• 56-64 kbits-per-second.  This does a pretty good job even for music.  Most people can't hear the "watery" or "gravelly" sound unless they know what they are listening for.  You might get away with using this as one channel of a backup on an FM radio station...  Achievable frequency response (which you trade for the "gravelly-ness") is pretty good.  (Monophonic or Joint Stereo)
• 128 kbits-per-second.  You are into the point of diminishing returns.  You reduce the "gravelly-ness" even more even with good frequency response.  Most people would be hard pressed to hear the artifacts in this unless they knew exactly what they were listening for.    This is still not quite CD-quality in my opinion.  (Joint Stereo)

Keep in mind that the above ratings are my own opinion based on the equipment that I have used.  Your mileage may vary.

In the real world, the data rates aren't always fixed.  In the case of Digital Satellite Television, several different programs are lumped together on one faster data stream.  There might be two news channels (total of 4 Mbps) plus two network channels with, say, sitcoms from the 60's (add two 3 Mbps for a total of 10 megabits-per-second) plus two sports network (add two 6 megabit-per-second channels) for a total of 22 megabits-per-second.  Let's say that the audio for all of these adds another 1 megabit-per-second, so the entire thing runs at 23 megabits-per-second.  You could get away with running these all at, say, 18 megabits-per-second, though.  Why?  Well, the time you need the high bandwidth is when you have lots of changes in your video.  When nothing much is happening (which is most of the time, when you think about it) you don't really need to be screaming along at full speed.  The likelihood of all of the channels being really busy all at the same time is pretty low and even if they do get busy, they aren't likely to stay busy during the entire time.  If the sports channels were both busy, they could momentarily steal some bandwidth from the news and network channels (people won't really notice a brief increase in the artifacts while they are engrossed in the TV show...)  This technique is called Statistical Multiplexing and is a neat trick to save even more bandwidth.

How is this transmitted?

In the case of satellite television, QPSK (Quadrature Phase Shift Keying) is usually used.  Because the signals from satellite are so weak, a fairly robust means of modulation (i.e. putting the information on a signal) is used.  For QPSK, signals need be only a little ways above the noise and do a pretty good job.  This includes, of course, error correction, too.

Digital Satellite television has the added advantage that either you can see the satellite, or you can not!  If there is anything blocking the clear shot into the portion of the sky where the satellite is, you are simply out of luck.  End of story.  If you can see the satellite, then that's because nothing is in your way.  (If you are looking through trees, then that's your problem, as you might already know...)  On the satellite, the signals are weak, but since you have a highly directive antenna pointing at the satellite, you won't be bothered by another satellite in another part of the sky operating on the same frequencies (channels) that the one you are looking at is on, so the number of and width of the channels you have available is less restricted (because another satellite can re-use channels, remember?)

Terrestrial (ground based) Digital television is different for several reasons:  The frequencies available for satellite are reserved for satellite only and are not available on the ground.  There is also a strong impetus to use frequencies that are already available (unused TV channels!)  This allows some adaptation of equipment and techniques long-established to be used, and (supposedly) allows existing mass-produced technology to be more easily adapted for the consumer market.

But all is not well.  The 18 megabit channel I exampled above takes about 27 Megahertz of bandwidth (more or less) to transmit - that's almost five television channels worth!  It doesn't take a genius to figure out that in a metro area, there aren't going to be enough unused channels available!  (Keep in mind that you can't use the channels that they use in the surrounding cities, either!)

What to do?  Well, the modulation scheme is different.  On satellite, they pick a modulation that carries well when the signals are weak, where on the ground, they optimize it for bandwidth.  So, they can cram into one television channel more than that 18 megabits worth of data.

All is not well, though.

The schemes that have been chosen are, as I said, optimized for getting more data into a smaller space.  Doing that has some tradeoffs:  The signal needs to be stronger and more pristine.  That is, the terrestrial signal is much more susceptible to distortion (due to transmitter problems, local interference from computers, power lines, reflections, etc.) than its satellite counterpart.  If you have a digital satellite television receiver you'll know that either you get a good picture, or you don't:  There's nothing in between.  Let's take a brief look at some of the problems that are (will be) experienced with terrestrial (ground based) digital TV:

• Transmitter problems:  The ground based transmitter must be working perfectly in order to transmit the signal without damaging it.  With analog TV, if the transmitter was having problems, it would show up as noise, brightness problems, color problems, or a weak signal.  While the signal might degraded, it could still be very watchable.  From a technical standpoint, an analog TV transmitter must suffer some very serious degradation before the average viewer will notice that anything is wrong.  For digital signals, distortions of magnitudes that wouldn't even be noticed by the average viewer on an analog signal may obliterate the digital signal to the point where it may not be able to be received.  This isn't as bad as it would seem, though, because transmitters are better than they ever were, and the broadcaster will likely have some sort of backup in place.
• Interference.  The nature of the frequencies used by satellites makes them nearly immune to things like power line noise, and interference from things like local transmitters and computers;  the highly directional antenna that is the dish helps prevent interference, too.  The frequencies used for television are much more susceptible.  Even though error correction is used, even slight interference is enough to wipe out a signal.
• Multipath.  (Also known as ghosting.)  This, I believe, will be the nemesis of ground based digital TV.  Ghosting caused by reflections from buildings, mountains, passing cars and airplanes can utterly trash the received signal.  If you are trying to use an indoor antenna (like a pair of rabbit ears) you'll be bitterly disappointed:  The signal will likely drop in and out just by someone moving anywhere nearby.  An indoor antenna is also much more susceptible to interference since it is likely physically much closer to the source of the interference, and the surrounding building weakens the desired signal anyway.
• Another problem is that with the "you have it, or you don't" nature of digital signals, it is much harder to diagnose exactly what is wrong.  With an analog TV signal you can just look at it and diagnose most types of problems.  That will not likely be true with digital TV...  at least, at first...  (That's assuming that someone would know how to interpret the diagnostic information anyway, but that's another issue...)

What does all of this mean?  I'll summarize everything up in brief, and make still more comments:

• Digital Television signals are inferior to Analog ones.  At least presently.  If you can cleanly receive that analog signal, it will look and sound better than the digital one, largely because of the compression artifacts.  Keep in mind that many of the signals you see on digital satellite are simply digital retransmissions of analog ones that the satellite uplink receives.  Practically speaking, they are more likely to have a better receiver and antenna than you would, so their signal looks better than one that you would likely get without going through a lot of trouble.  Keep in mind that there are very few HDTV programs available at the moment:  Those that are transmitted will probably look better than their analog counterparts.
• Terrestrial digital television signal are very fragile.  You will probably not be able to get away with indoor antennas at all in most situations, and even if you use an outdoor antenna, your signals will have to be relatively free of multipath, interference, and be strong enough.  If you live in a big city with a lot of buildings, you may be in big trouble.
• The digital signals are not forgiving.  With analog signals, just looking at them can often tell you what is wrong:  If it's snowy, it's weak.  If it's ghosty, then re-orient the antenna.  If there are lines, then you have an interference source.  Digital signals will either be received, or they will not.  Period.  Making adjustments to digital signals will be a pain since, unlike an analog receiver, you may not know instantly if what you did made things better or worse:  The digital receiver may take a brief moment to re-lock on the signal.  Even then, it may not be able to really tell you if you improved it by a large amount, or are on the hairy edge of losing the signal again.
• Remember that Digital TV does not mean High Definition TV.  There are reports of so-called experts at electronics stores telling people otherwise.  There is even mis-information being doled out at some of the larger electronic retailers that lead consumers to believe that an S-Video connector is used for HDTV right now, or that the presence of an S-Video connector guarantees HDTV compatibility.  It does not!
• When digital TV appears, your analog set will not be useless:  Analog signals will be broadcast for quite a while to come (especially if there are problems...) and there are already adapters that receive digital signals and convert them into a form that your old TV can use.
• One digital TV signal can carry multiple programs.  Your local affiliate may have an HDTV signal or several lower-resolution network programs (maybe the same network but with the programs time delayed, so you can watch them at your convenience) or even some pay-per-view channels.
• In large geographical areas (such as the western U.S.) where most of the land area is served by low power TV translators, it is unlikely that the television stations will soon wish to invest in putting digital capable facilities on the air to serve sparsely-populated areas.  In many areas, it is local (city our county) government that (barely) supports the local translators.  The large expense for new equipment and the hugely increased technical requirements for transmitting a digital signal will make maintenance of these systems more of a nightmare than they currently are.  Most likely, these areas will simply be left out, having to rely on the old analog systems (while they exist) or resorting to satellite reception.  (This is already happening in rural Utah!)

For some additional information about Digital Television, go to the North American MPEG Information page.

This page last updated 20000809