Bad Encoder Settings on HD AM?

[FTL_Ian]

So far, I have listened to two different stations in HD on AM: WZBK and WBZ

ZBK is a "music of your life" station and it sounds pretty good until the commercial breaks. It is at this point that one particular compression artifact becomes very noticeable. I'm not very well-versed in the technical terms for what it is I'm hearing, but it is very obvious on words that begin or end with "s". (Is the term for this, "phasey"?)

At one time, I had experienced a similar artifact when I was compressing my talk show into a 32 kbps MP3 file. I solved the problem by adjusting the maximum bandwidth in the compression settings from 7634 Hz to 6000 Hz.

I figured that since ZBK is a music station that they must have their maximum bandwidth set pretty high on their encoder and that is why the artifact is present on voices.

This weekend though, I heard the same exact problem on WBZ. I figured that an all talk station would have their compression settings tweaked for voice. Apparently not.

Are stations even able to adjust these settings?

Has anyone else heard what I'm describing and does it bother you too?

 

[LinoNYC]

Has anyone else heard what I'm describing and does it bother you too?

Yep although I complained about this from the late '80s ' it is worse w/iboc.

Alot of commercials are sent to stations as Mp3 or other compressed, network feeds are also compressed, add to that station's use digital obscenity delay and some still use compressed STLs, then this all gets fed to the iboc unit which compresses it again. The effect is called cross coding it was a problem long before iboc which just made it worse.

Lino

 

[JohnnyElectron]

I hear it on a local AM News/Talk too - nasty sounding 'sssss' (siblience) on every spot and even on 'live' call-in shows (okay, plus potty-mouth delay) - but I don't hear it on analog - only on the HD-AM. That's why I wish that they had an 'analog-only' 'switch' on the DEI HD tuner, as analog sounds better than HD-AM 80% of the time on the N/T station.

The AM sports station doesn't have as bad of an 'sssss' sound on HD, but the telephone calls sound horrible and 'fade' into the background - almost like an 'out of phase' audio signal. Music sounds great, but the level of the talk and calls is at least -3dB of the music bumpers. Strange; but then again, the way the AGC works (or doesn't work) on my Directed Electronics CAR IBOC HD tuner, the audio level gets cranked too low, it goes into HD and blows the speakers.

 

[R.F. Burns]

I hear it on a local AM News/Talk too - nasty sounding 'sssss' (siblience) on every spot and even on 'live' call-in shows (okay, plus potty-mouth delay) - but I don't hear it on analog - only on the HD-AM. That's why I wish that they had an 'analog-only' 'switch' on the DEI HD tuner, as analog sounds better than HD-AM 80% of the time on the N/T station.

The AM sports station doesn't have as bad of an 'sssss' sound on HD, but the telephone calls sound horrible and 'fade' into the background - almost like an 'out of phase' audio signal. Music sounds great, but the level of the talk and calls is at least -3dB of the music bumpers. Strange; but then again, the way the AGC works (or doesn't work) on my Directed Electronics CAR IBOC HD tuner, the audio level gets cranked too low, it goes into HD and blows the speakers.

It isn't that analog AM sounds better then IBOC any more than a used LP sounds better than a CD. What you are experiencing is the effect of how much more revealing IBOC is than restricted analog AM audio. Due to the other issues inherent with analog AM (Resticted frequency response, poor signal to noise, relatively poor dynamic range) it isn't that these audio issues don't exist due to analog transmission methods. Its because all of the other problems which AM is effected by are covered up by the AM receiver's lack of faithfully reproducing truly wide band (20 Hz to 20Khz) audio with a s/n of over 80 db.
Listen to the lead in and lead out grooves of a analog LP and listen to groove wall distortion. That kind of noise was filtered out by our ears for many years, but when presented with crystal clear digital audio all of the hiss and groove noise associated with vinyl became very apparent. Even the early CD's which were transfered from analog source material commented that hiss will be heard on a recording due to the source material. Cd's of course have improved over the years as has every technology ever devised.

 

[LinoNYC]

What you are experiencing is the effect of how much more revealing IBOC is than restricted analog AM audio

True, to an extent. The available data bandwidth of a AM iboc is 36K, I have seen a spectral analysis , "pure" digital exists up to approx 4.5khz above that it is all spectral replication. Just like MIDI synthesis SBR is dependent on codec quality with the added complication of requiring carefull preconditioning of the audio. Among others it's best to avoid top-heavy mixes.

Lino

 

[R.F. Burns]

What you are experiencing is the effect of how much more revealing IBOC is than restricted analog AM audio

True, to an extent. The available data bandwidth of a AM iboc is 36K, I have seen a spectral analysis , "pure" digital exists up to approx 4.5khz above that it is all spectral replication. Just like MIDI synthesis SBR is dependent on codec quality with the added complication of requiring carefull preconditioning of the audio. Among others it's best to avoid top-heavy mixes.

Lino

When I say revealing, I refer to digitals far superior signal to noise and dynamic range numbers , both of which are much better then what is available over analog AM radio.

 

[scanman1]

I have a couple of stations in my area that have issues with "siblance", but it's not bad enough to make me tune out. I prefer the digital (when I can get it to stay locked in).

 

[Mike Walker]

Now I love RF...he's my bud. And what he says about the superiority of digital in terms of signal to noise ratio and frequency response is obvious. But digital artifacts ARE a problem at low bitrates. Some people are less sensitive to this phenomenon than others. And there are ways for stations to minimize it. But there's a reason a 96kbps stream sounds better than a 32kbps stream (AM HD is 32, not 36kbps). There is more data available to encode the signal.

Lossy codecs split the audio spectrum into various bands, and then decide moment to moment which of these bands it is critical to encode, and which can be ignored without audible consequence. Today's codecs do remarkably well. For instance, the HDC codec used in HD Radio can be as transparent at 96kbps, or even 48kbps as older codecs at much higher bitrates. But 32kbps IS low. What some describe as "phasiness" is actually a comb filter effect. Your ear/brain is telling you that the audio spectrum has been "sliced and diced" into tiny pieces, and not only are some of those "slices" missing, but the ones missing now are completely different than those missing a second or so ago...which causes the "swishy" sound on high frequency sounds such as cymbals and sibiants ("s" and "t" sounds, for instance).

This effect is known, and if you think about how lossy compression works, the fact that it becomes worse at low bitrates is a no-brainer. But all is not lost...low bitrate encoding CAN sound just fine, if engineers at the station are using proper pre-processing (which pre-conditions incoming audio to avoid artifacts), and if they don't go crazy with dynamic-range compression, and extreme levels of high frequency content. If you hear artifacts, call the station and yell at them! They need to know that we CAN hear these artifacts, and aren't the least bit happy about it. They'll learn in time. This technology, especially on AM, is brand new. Everybody's still learning.

 

[R.F. Burns]

Now I love RF...he's my bud. And what he says about the superiority of digital in terms of signal to noise ratio and frequency response is obvious. But digital artifacts ARE a problem at low bitrates. Some people are less sensitive to this phenomenon than others. And there are ways for stations to minimize it. But there's a reason a 96kbps stream sounds better than a 32kbps stream (AM HD is 32, not 36kbps). There is more data available to encode the signal.

Lossy codecs split the audio spectrum into various bands, and then decide moment to moment which of these bands it is critical to encode, and which can be ignored without audible consequence. Today's codecs do remarkably well. For instance, the HDC codec used in HD Radio can be as transparent at 96kbps, or even 48kbps as older codecs at much higher bitrates. But 32kbps IS low. What some describe as "phasiness" is actually a comb filter effect. Your ear/brain is telling you that the audio spectrum has been "sliced and diced" into tiny pieces, and not only are some of those "slices" missing, but the ones missing now are completely different than those missing a second or so ago...which causes the "swishy" sound on high frequency sounds such as cymbals and sibiants ("s" and "t" sounds, for instance).

This effect is known, and if you think about how lossy compression works, the fact that it becomes worse at low bitrates is a no-brainer. But all is not lost...low bitrate encoding CAN sound just fine, if engineers at the station are using proper pre-processing (which pre-conditions incoming audio to avoid artifacts), and if they don't go crazy with dynamic-range compression, and extreme levels of high frequency content. If you hear artifacts, call the station and yell at them! They need to know that we CAN hear these artifacts, and aren't the least bit happy about it. They'll learn in time. This technology, especially on AM, is brand new. Everybody's still learning.

And you and I are on the same page Mike. I know all about cascading codecs. That's why we've gone to linear T's from compressed T's. We're moving from MP2 servers to wav file servers too. All that aside there are so many levels of compression prior to local station transmission that it's near impossible at this time to keep all artifacts out of the chain. I've said before I am not impressed with HD on AM but not for the reason many of the anti IBOC people are. I am a lifelong DXer but I don't care one bit about DXing if AM audio can be improved. First adjacent interference? The fact that I can't hear WLW in NYC on the radio any longer? Big deal. The small station trying to get an audience in a non protected area (In my area a 1K foreign language station trying to reach Brooklyn with a good signal and Brooklyn is over 30 miles away). Try servicing your real local market with compelling programing and if there isn't enough business in your local market to keep the station profitable then maybe the station should be moved or not on the air at all. AM Iboc sounds great on my Receptor but when played throuogh a real high end system the limited bit rate really shows through.

 

[Play Freebird]

And you and I are on the same page Mike. I know all about cascading codecs. That's why we've gone to linear T's from compressed T's. We're moving from MP2 servers to wav file servers too. All that aside there are so many levels of compression prior to local station transmission that it's near impossible at this time to keep all artifacts out of the chain.

But why are AM digital artifacts still so evident on live voice? This is source material that has never touched a lossy codec, other than the HD encoder and decoder. I sure notice them on KYW 1060.

Also, if your compressed T's are using APT-X, there's little to be gained by going to linear T's, unless the extra bandwidth on the line is available at no additional cost. APT isn't a perceptual (lossy) codec, it's just a refined form of "Delta Modulation" in which the difference between successive PDM samples is coded, rather than the absolute value of each sample. Therefore, it's quite compatible with cascaded lossy codecs.

 

[Ray22]

For AM HD: That's 32kb for the STEREO data! You figure out what this means when you listen to AM HD. 16kb
To my ear it sounds like the worst ISDN remote you have ever heard, more like a bad phone line codec sound but with more phasing in and out sound. The music or spots not real bad but on talk formats it's pretty bad especially if driving. This is close in to transmitter sites (several) 20 miles or less.

 

[Mike Walker]

Digital artifacts are still audible on live voice because, and many will yell at me for saying this (I'd propose they LISTEN WITH THEIR EARS, rather than their pre-conceived notions) it's harder for a lossy codec to encode voice by itself. Think about it...with lossy compression, something is always missing. "Successful" encoding is about the codec "hiding it's tracks". Music's complexity leaves lots of room for some sounds to mask others, so that by the time you just about had a handle on an audible artifact, it's gone (more precisely changed, as the codec adapts once again to a changing signal). But with spoken voice, there's no place to hide! First of all, we are more familiar with the sound of human voice than any musical instrument...t's the "instrument" God tuned our ears to be most sensitive to. If something's "missing" from spoken word, it's INSTANTLY apparent. Which is why stations have to be VERY careful about too much dynamic range compression/limiting on the HD stream of talk formats. Pushing up the average level makes the codec work harder, and gives it fewer options to "hide it's tracks". Now reduce the bitrate as low as it is on AM HD, and the poor codec is SCREAMING "UNCLE"! Something's gotta' give, and that something is fidelity. This is not an unsolvable problem. If AM engineers just give up the LOUDER IS BETTER philosophy they've lived-by for decades, and backed the hell off on audio processing, the world would be much brighter and sunnier!

 

[Play Freebird]

If AM engineers just give up the LOUDER IS BETTER philosophy they've lived-by for decades, and backed the hell off on audio processing, the world would be much brighter and sunnier!

Therein lies one of the major challenges of IBOC broadcasting; we must maintain similar levels of processing on both the analog and digital channels so that the change in level isn't too objectionable as the receiver "blends" back and forth. Few engineers want to back off their AM analog processing because the medium-wave spectrum is naturally and artificially noisy -- even more so since digital hash was introduced to the band.

Obviously, for best results you should avoid clipping ahead of a lossy codec -- and there are some other tricks that reduce artifacts -- but you are correct in realizing that the available bitrate on the AM digital channel is simply inadequate for high-fidelity audio.

 

[FTL_Ian]

For AM HD: That's 32kb for the STEREO data! You figure out what this means when you listen to AM HD. 16kb

Stations do have the option to transmit in Mono HD, do they not? An AM talk station would be insane to run in stereo.

 

[Mike Walker]

I don't know if it's "insane" to broadcast talk in stereo, FLT_Ian. In the UK, where technical standards embarrass most American broadcasters, MOST talk (or "speech" as they call it) is broadcast in stereo. WE HEAR SOUND FROM MORE THAN ONE DIRECTION! The depth and dimensionality of stereo, WHEN USED PROPERLY, can greatly enhance non-music programming!

How does football or a Nascar race sound through your home theater rig, on HDTV? THAT AIN'T MUSIC! Especially with digital's arrival, I think American broadcasters should get over the idea that stereo is only for music. In the early days of stereo recording, AND BROADCASTING, people were a lot more creative! What's "insane" is heavily compressing and clipping an audio stream that's going to be encoded at a low bitrate. It's asking for trouble. But making use of the new technology in all it's glory is far from insane! If we're going to settle for "mono/speech quality/limited bandwidth", then HD isn't worth the compromises being made on AM! But if we're going to use the technology to SURPRISE listeners at what they hear...well that's a different can of, er, decibels

 

[FTL_Ian]

I don't know if it's "insane" to broadcast talk in stereo, FLT_Ian. In the UK, where technical standards embarrass most American broadcasters, MOST talk (or "speech" as they call it) is broadcast in stereo. WE HEAR SOUND FROM MORE THAN ONE DIRECTION! The depth and dimensionality of stereo, WHEN USED PROPERLY, can greatly enhance non-music programming!

It is indeed insane to encode talk in stereo when you have limited bandwidth as AM HD stations have.

As a listener, all I care about is what sounds good. Stereo at 32kbps sounds worse than Mono at 32kbps - period.

 

[Mike Walker]

It isn't so black and white, and it isn't that rigid. Stereo at 32kbps CAN sound bad, BECAUSE aggressive audio processing and clipping produces artifacts in the codec. Back off on the audio processing, and use proper pre-processing, and the artifacts subside. Noise-free stereo is one of the greatest benefits to AM stations of this technology! It shouldn't be dismissed!

Besides, as a production guy I can tell you another benefit. The sponsor's messages...those things that pay the freakin' bills...sound much better, and allow for much more creative options IN STEREO! So even if everything's "dead center" during the talk segments (except the bumper music), the spots can actually be more effective in stereo! And with pocket-sized high resolution digital recorders costing less than the Marantz cassette machines those of us in radio are so familiar with, capturing news actualities in stereo is a no brainer...bringing a remarkable sense of "being there" to the news. And BEING THERE is kind of the point of broadcast news coverage. Don't dismiss one of the greatest creative tools in broadcasting!

Early FM stereo had a bumpy start. Remember the (moronic!) using TWO microphones for the announcer? People who listened in stereo got seasick, as the announcer moved back and forth across the room as he turned his head to cough, and mono listeners got a hollow, phasey mess. We learn! We advance! 32kbps HDC today can sound as good as the average 128kbps mp3 did a decade ago (mp3 encoding has improved too...and it's a VERY old, inefficient codec!) We'll learn, we'll stumble, and we'll eventually get it right. Early Technicolor sure as hell wasn't accurate. It captured less resolution than black and white (by a long shot!) But aren't you glad cinematographers didn't give up on color because "after all, drama is mostly people talking, and current technology is good enough"?

 

[dbdigital]

Early Technicolor sure as hell wasn't accurate. It captured less resolution than black and white (by a long shot!) But aren't you glad cinematographers didn't give up on color because "after all, drama is mostly people talking, and current technology is good enough"?

Actually since Technicolor used three strips of black and white film, it did capture all of the resolution (as the lenses of those times allowed). The problem was the accuracy of the registration of those strips as they were printed onto a single film.

It hasn't been until recently with the advent of digital film scanning that we've seen truly accurate registration of Technicolor film matrices and have seen just how sharp films shot in this process really were. They're actually sharper than films shot on "sandwiched" or conventional color film since the light tends to scatter as it penetrates the layers. So, in a way, we've actually gone backward as far as color film processes is concerned.

But I get your point.

db

 

[Mike Walker]

By "inaccurate" I was talking about color reproduction. There's a reason Technicolor required that studios employ a "Technicolor Color consultant". The colors you got with Technicolor were beautiful, but they didn't look anything like the actual colors that were present on the set! You're right though about reigistration issues with 3-strip Technicolor. Still, three-strip Technicolor was beautiful! But you're right, Technicolor may have (we find now) bee sharper, but it had less accurate color reproduction than today's color film. So what? Filmmaking ain't about accuracy...it's about getting a "look". Kind of like recording (except for certain acoustic music where the goal IS accuracy).

In general, black and white photography still captures greater resolution, and dynamic range (range from black through all the shades of grey to white) than color. Hey dbdigital, I LOVE talking about old filmmaking, and photography! It's weird, I'm legally blind (retinitis pigmentosa), but I'm fascinated by photography. Always have been! My wife still prefers to use a film camera. I keep trying to convince her that today's digital cameras are actually sharper!

 

[dbdigital]

By "inaccurate" I was talking about color reproduction. There's a reason Technicolor required that studios employ a "Technicolor Color consultant". The colors you got with Technicolor were beautiful, but they didn't look anything like the actual colors that were present on the set! You're right though about reigistration issues with 3-strip Technicolor. Still, three-strip Technicolor was beautiful! But you're right, Technicolor may have (we find now) bee sharper, but it had less accurate color reproduction than today's color film. So what? Filmmaking ain't about accuracy...it's about getting a "look". Kind of like recording (except for certain acoustic music where the goal IS accuracy).

In general, black and white photography still captures greater resolution, and dynamic range (range from black through all the shades of grey to white) than color. Hey dbdigital, I LOVE talking about old filmmaking, and photography! It's weird, I'm legally blind (retinitis pigmentosa), but I'm fascinated by photography. Always have been! My wife still prefers to use a film camera. I keep trying to convince her that today's digital cameras are actually sharper!

Mike:

Without getting into a discussion on the merits of film verses digital photography on, of all things, a radio board, I will say that I use both for my work depending on a client's needs. Each has their advantages for certain applications.

However, if I'm shooting something really special, something that requires archiving, I will use film. So I agree with your wife.

You can see some of my work here.

http://www.danielbrownproductions.com/

If you haven't already seen this film, check out Orson Welle's "It's All True". One of the segments is a Mardi Gras in Brazil (I believe) that Welles shot in Technicolor. Welles was attempting to use Technicolor in a documentary setting, away from the controlled confines of a studio. The color is vivid but surprisingly accurate. The only problem I noticed (in a few shots) was that if the light fell below a certain footcandle the image completely disappeared; no murky outlines, nothing. So Technicolor must have had a very narrow slope unlike today's color films which allow for some underexposure.

Sorry, back to a discussion on HD Radio.

db