ATSC Digital Audio Quality, Technology, Issues

[JamesK]

^^^^

I did not confabulate anything. I gave two examples of error correction. Broadcasters would use forward error correction, which does not require retransmission. I also never said digital is error free. I pointed out that error detection and correction are commonly used to ensure data integrity. I also pointed out that it is possible to overwhelm forward error correction. If digital were error free then there'd be no need for error detection & correction. On the other hand, analog has no mechanism to recover from the noise and distortion that are introduced at every step of the process. If you send analog signals over a noisy or distorted connection, all that noise and distortion accumulates. With digital, you can sustain a significant amount of noise and distortion and still recover an error free signal.

[holl_ands]

ATSC A_53 defines Audio specs (16 bits/sample min & 24 bits/sample max, both at 48 ksps).
Most of the Pro Audio equipment used in the past 5+ years has employed 20 or 24 bit sampling:
http://www.atsc.org/standards/
There can be several DD5.1 audio streams, each in a different language. Also note Visually
Impaired and other data streams that can be carried along with the primary audio stream.

ATSC A_52B is Dolby's general purpose specification for AC-3, some of which does not apply to ATSC.
The introduction sketches out how the very high input data rate across all 5.1 channels is compressed
into just 384 or 448 kbps for the transmitted AC-3 data stream. Stereo can be a much lower data rate,
Laser Disk is fixed at 384 kbps and on Blu-Ray it can be as high as 640 kbs.
[Each channel is 24 bits@48 ksps=1152 kbps or 768 kbps for 16 bits.]
http://www.extremetech.com/article2/...1156782,00.asp
http://www.dolby.com/uploadedFiles/z...l/42_DDFAQ.pdf
http://www.dolby.com/DocLibTechLanding.aspx?taxid=584

"Perceptual Coding" starts by performing a Fourier Transform to determine the signal strength in
256 frequency bands. Some algorithms are used to encode the "frequency bin" strength and duration
for the loudest of these frequency bands, ignoring those that are "masked" by strong bins.

So it's easy to reproduce a single frequency test tone...."perfectly", i.e. less than 0.001 %
distortion per Dolby Encoder DP563 specs:
http://www.dolby.com/professional/pr...gic/dp563.html
Note the Rear Surrounds roll off at 7 kHz...that would be for matrixed Pro Logic mode only....not DD.
Low freq spec is 20 Hz. They state that upper frequency response depends on sampling frequency.
At 48 ksps, that would be not quite 24 kHz....higher than what nearly any person can hear.
For comparison CDs are 16 bit at 44.1 ksps (22 kHz) and most "hi-rez" audio (DTS, DVD-Audio, SACD)
are either 24 bits@48ksps or 24bits@96sps (Stereo-only to push the D/A converter's LPF even
higher in frequency to avoid (inaudible?) phase and time delay distortion at the highest frequencies).

The reproduced distortion and frequency response are primarily determined by your AVR, the
speakers and what your room is doing to the surround soundfield. It probably is somewhat worse
than the Dolby DP564 Decoder:
http://www.dolby.com/professional/pr...gic/dp564.html
Frequency Response is an ideal 20-20 kHz with distortion a whopping 0.02 %.
Note they don't describe the measurement conditions....for that there are a bunch of CEA standards
that you can buy for about $100 a pop....

Since the technical specs are so close to "perfect", they are almost never cited in most documents.

The major difference between perfection and what AC-3 (DD5.1) actually delivers is hidden in
what the Perceptual Encoding algorithm is doing to the overall sound field....the results of which
can only be "measured" by your ears (or rely on reviews by highly trained audiophiles):
http://elvera.nue.tu-berlin.de/files/0747Noll2005.pdf
http://www.eas.asu.edu/~spanias/pape...spanias-00.pdf
http://www.ece.rochester.edu/~gsharm...udioCoding.pdf

Which is why DTS (lossy like DD, but higher 1510 kbps data rate) and the newer loss-less
digital audio formats were invented, such as SACD, DVD-Audio and the new Blu-Ray formats.

CLASS DISMISSED!!!!

[stampeder]

bentoronto, the key principles for your reading are FEC and BER, which, as JamesK and holl_ands have been clarifying, are addressed by the ATSC standard so that errors do not require retransmission of data. Neither of them have suggested that perfect data transfer occurs, and indeed we've been discussing this issue recently in the following thread:

Debugging Momentary "Blips" in my ATSC Reception

Back to the audio discussion here.

[bentoronto]

Has anybody been saying the output will be perfect?

Only JamesK:

As long as you have sufficienct S/N ratio to properly recover the signal, the output will be "perfect".

OK, he said "'perfect'" not "perfect."

Back to the thread topic: what are the performance specs?

Seems like a pretty valid question.

But first this was dismissed because given enough bits, FEC, BER, etc. the transmission part of the process was pretty good or as some would say with profound confidence in the principles of engineering, perfect.

But even when that part of the process is pretty good, it is still fair to ask about all the pieces in the process and the total distortion, noise, bandwidth, etc. I am not asking to refer back to the piano tuner, but at least back to the broadcast console input and forward to my first RF stage.

Like with other hi-fi gear, certain definitions and standard tests need to be somewhat arbitrarily defined. But that's normal.

Even with elements of variable compression, variable bandwidth, noise reducers, etc. confusing things, I think consumers of the transmission sound quality (and manufacturers) deserve to have some yardstick of quality.

[JamesK]

^^^^
The bottom line is no matter how good or bad the source, the digital transmission system will reproduce it faithfully. That is any problems are not likely to occur in the digital stages. If the signal is poor enough, you may have audible hits, but it will not cause a bad sound mix. With the analog stages, you need things like flat frequency response, low noise & distortion etc, with things getting worse at every step.

[holl_ands]

There are some performance SPECS for CECB converter boxes:
http://www.ntia.doc.gov/dtvcoupon/dtvmanufacturers.pdf
http://www.ntia.doc.gov/frnotices/20...ule_031207.pdf
Note the use of 3x10-6 ber test criteria, 15.5 dB SNR against co-channel interference,
various adjacent channel requirements, Burst Noise tolerance and Field Ensemble tests:

They are a subset of the A/74 Receiver Performance "Guidelines" (which are NOT mandatory):
http://www.atsc.org/standards/practices.php
In addition to a bunch of mandatory specs:
http://www.atsc.org/standards/

[bentoronto]

Your efforts and links much appreciated.

I think there is a great divide between empiricist people (like me) who are looking for performance standards of performance end-to-end and those who work with operational parameters (for want of the right term - and I've never figured out what is the right term for the other kind of standard).

Although I am not able to read let alone understand all that prose, I was unable to find the term "distortion" anywhere. Likewise, "noise" was present in discussions but not kind of nailed down as a listener parameter, at least as far as I could tell.

[majortom]

I think ya may be just a little confused, from what I can gather, u sound like a hi-fi buff. You know, listen to CD's on the most expensive stereo system known to man, like it's gonna make all the difference in the world listening to Bach...When the rest of the population is perfectly happy with the tin can speakers in their car on their way to work jammin to Black Sabbath or sumthin clipping their their amp to the the dc rails. Nothin wrong or right with either if it floats ur boat, a matter of personal taste is all. Who cares?

Putting that BACH cd in ur cd player there is no nasty transmission medium to overcome, like in radio communications. So ur not concerned with the typical terminology of Radio or Television Transmission, or any radio communications environment Digital or Analog, makes no difference. The environment is the same. Same thing, only different. Same principles apply, You'll hear the same terms in both worlds, Receiver Sensitivity, Noise figure, Selectivity, Intermodulation, creates distortion (ya happy now) in radio signals...

With such audiophile yardsticks in hand, as a consumer, when is the last time you ran a Two-Tone Intermodulation Distortion test on your hi-fi Amp? And at what Power level did it start to clip? Did it even come close to the manufacturer's specification? What is the 1dB compression point in your Hi-FI Amplifier? If it did not come close, what would you do about it? Take it in for warranty service? Would you be able to get Warranty service? Would the guy perfoming said warranty service even run these types of tests after repairing it (not very many would)? Would you buy a different amp perhaps?

The same types of measurements in the design of Radio communications equipment, to make sure the signal gets to the other end properly . Do you run around making these measurements every day or do you care? Probably not. Bottom line, at the end of the day, can you understand the intelligence in the signal that was transmitted to you over the air? Yes. If the OTA TV PIX and sound is not to your satisfaction, then you may need to buy a different set, get a new antenna system, or even subscribe to cable if that works better for you. Just like ya would just go buy a new amp if ya didn't like ur present hi-fi Amp.

There is plenty of information available in the numerous threads here for anyone to make an intelligent decision on what to buy or not buy when making such purchases, when it comes Hi Def, Standard def, Antenna Systems, etc. Whatever floats ur boat... it can be found here.

[JamesK]

^^^^
Actually, many years ago, I'd do noise & distortion tests once a week on the CBC radio feed to some low power transmitter sites in Northern Ontario. Every Thursday afternoon at 1:00, the CBC would send the various test signals down the line, to measure things like frequency response, noise level, harmonic distortion etc. I'd make the measurements and send the results back to the CBC.

[majortom]

many years ago...

See how much things have improved? Not necessary anymore...rest my case.
hehe...Just kidding. That sounds cool. Did ya work for them or sumthin?

[JamesK]

^^^^
I worked for CN Telecommunications, which became CNCP Telecommunications, which became Unitel Communications, which went bust. Nowadays, remote testing is the norm. However, my experience has long shown me the benefits of digital systems over analog.

[bentoronto]

Yes, majortom, I am interested in specs for a whole bunch or reasons and at different times in my cycle of home-listening development, esp. when I'm thinking about a $700 upgrade (HD FM tuner, antenna, installation, etc.).

It seems beyond my ability to explain in this thread that I want to know (1) "what specs of any sort will help illuminate if I will be happy with the sound quality X-system delivers to my listening room?" and NOT (2) "how many bits get lost in the ether during the part of the process which could be narrowly called the transmission link?".

(1) is a worthy question and I don't think much of this thread has addressed it.

[iblackford]

if you really aren't concerned with (2) the transmission link, then the only things you need be concerned with are what happens at the source, which holl_ands post #29 in this thread has answered fairly well.

My TV at least will output the pcm audio stream so that my receiver can decode it, you should definately look for a TV/receiver with this feature, I think.

If you are still aren't satisfied, I suggest you contact the broadcaster or broadcasters in question, to find out what equipment they are using prior to digitization.

Ivan

[stampeder]

bentoronto, I don't see much progress within this thread towards satisfying your curiosity. I am gently suggesting that the answers you are seeking might not be found here amongst us.

[chengbin]

Anybody notice that OTA HD's 5.1 sound lacks bass compared to analog broadcasts?

I've noticed it before when watching talent shows, and their hip hop music is lacking bass. I thought it was just the way it is mixed so the bass doesn't overpower the performer, but I think it is something else. I didn't notice it until today, where I did a comparison by switching back and forth.

Today I was watching OMNI 2 in OTA HD, and it was playing some foreign music. I felt like switching to fuzzy SD just to see how bad it is, and I noticed the analog broadcast's sound has MUCH more bass. The ATSC's bass was almost non existent, while the NTSC's bass was punchy, the way it should sound.

Anybody notice that? Is there a reason?

Thanks.