Here's a sample (recorded from my Accurian HD Radio, patched into an Archos Gmini 402 hard drive digital recorder, captured in uncompressed .wav format, transferred to the computer in my studio, edited for brevity and emphasis, with my comments added, compressed to a high bitrate Windows Media file. Sorry, no lossless. I can only hold so many huge files at my website. The antenna was the Magnum Dynalab SR100 "Silver Ribbon", a glorified set of rabbit ears. Distant location. Simple INDOOR antenna. You be the judge of whether HD Radio is reliable at a distance, or merely a "farce" as others have claimed! The file may either stream, or be downloaded on a broadband connection. May the "farce" be with you!

http://www.theproductionroom.net/hd.wma

Nice work, Mike! Well if anything, I have to say that the HD audio I heard can keep up with anything I've heard on FM analog. You must have a nice antenna setup for what goes around this way as DX on FM. A nice benefit to the HD audio recovery is that unless you hadn't mentioned air-mile distances of the stations you recorded, there would be no discernable way of knowing the depth RF pathways of the station signals to your Accurian. I agree that it is a rather amazing set for an initial portable form factor effort, let alone something that could be distributed by Radio Shack. I enjoy its overall selectivity on both AM and FM.
I whole-heartedly agree with you, that all in all, the audio quality is quite acceptable given the technological infrastructure Ibiquity had to work with and you've demonstrated the DX-worthy virtues of listening to relatively long distance HD FM signals. Thanks for sharing, Mike
AW

Just superb, Mike. Thanks for the HD demonstration and overview.

The first capture of WDAV, the classical station (96k) sounded good, full, wide, and a little louder. However there were a couple of sparkles when the announcer was speaking. I wonder where that was coming from?

The second capture you recorded WTQR, with HD1, and HD2 sounded almost as good as the classical station. Its hard to tell when you are comparing classical selections to country or pop music. Really good fidelity. Wide, noticeable stereo separation.

Last station recorded (Lite) sounded compressed. no dynamic range. The highs were shrill, and squeaky. Good stereo separation on Marvin Gaye's (I heard it through the grapevine)

All were wide, as far as stereo presence is concerned. Good separation. My ears, my personal response. No test equipment. Well you asked, I'm ready to take the heat.

I have not idea what you mean by "sparkles". Perhaps you could point out where (in minutes/seconds) in the file this happens, and on what words/syllables. To me, through headphones, it sounds like the announcer (on WDAV) IS IN MY HEAD. VERY natural. I've heard lots of audio 'straight off the board'. THIS IS WHAT IT SOUNDS LIKE!

WTQR is loud and in your face BECAUSE it's processed TO HELL. Open the file in a program like Adobe Audition. THE METERS STAND STILL AND DANCE A LITTLE. That may sound impressive, but THAT is what compression sounds like NOT Lite 102.9. Lite 102.9 has less "presence" and is less loud BECAUSE THEY USE LESS AUDIO COMPRESSION/LIMITING (as opposed to data "compression"). There ARE artifacts from the lower bitrate on the HD2 streams, but they're certainly acceptable to these ears, particularly in comparison to what analog would sound like at EIGHTY MILES (hissy with multipath distortion, OR blended almost completely to mono).

This technology (ON FM) is the real thing. It delivers the goods. PRISTINE audio on a single stream, or pretty damn good audio with two streams. I'm down wit' that!

That may sound impressive, but THAT is what compression sounds like NOT Lite 102.9. Lite 102.9 has less "presence" and is less loud BECAUSE THEY USE LESS AUDIO COMPRESSION/LIMITING (as opposed to data "compression").
YES COMPRESSION SOUNDS LIKE THEY'RE TYPING IN CAPITAL LETTERS ALL THE TIME! ;)

Touche' Scowl! You zinged me, AND made an excellent point. That's exactly what it sounds like.

I knew that I would get hammered for that remark.The best way that I can describe the sound of a sparkle is a short high pitch sizzle sound at the loudest portion of the female announcers modulation. I call it a sparkle because it is only a millisecond in length. I hope that no one else can hear it. Then we can attribute it to my speakers, amplification or whatever.

Don't get me wrong, I have a Sangean HDT-1. I think the HD sound is great. I thought that you would certainly remark on my claim that your third sample was narrow, not wide, as far as dynamic range is concerned.

I can hear the sparkle my computer, and it's even more apparent when played back through my stereo and iPod.

I've actually figured out what it is - the compression. Yea, I know that's pretty much a DUH, but it's pretty interesting to watch the actual waveform (open it in an audio editor like Audacity or Adobe Audition and zoom in until you can see the individual samples, then hit play).

You'll see what the compression does.

Later today, I'll be posting a link to a web page with a collection of screenshots I took of the waveforms. I think it's pretty obvious what the 'sparkle' looks like. [basically, the compression "squares" some "rounded edges" creating that nasty, grainy sound.

"Squares" as in clipping are far more likely the station's limiter, not compression artifacts. Sorry. EVERY radio station uses some limiting...even digital ones. Or if they simply overdrive a component somewhere in the chain, then the waves got "squared". Look how close they go to 0dbfs on a regular basis. There is NO HEADROOM with digital media. Zip. 0dbfs is it. Overload anything in the chain, and waves get "squared". I observe (on casual observation) audio coming within .3db of 0dbfs on a pretty regular basis during the loudest part of the music. By the way, I don't hear "sparkles" (whatever that means). I DO hear much more extension than analog FM. How can I put this gracefully...some soundcards have poor DACs (digital to analog converters) that distort on the 16th bit...as audio approaches 0dbfs (remember each "bit" handles 6db of audio). This is why when sending work to radio stations, I always lower the level so that NOTHING peaks higher than -1db. I didn't do that with this file. I suspect if I lowered everything by 7db or so, which completely removes the 16th bit from the equation, the "sparkles" on your system would go away.

Want to know why it's almost definitely NOT the codec "squaring the waves"? Because a square wave is more complex than a sine wave (the simplest of all waveforms, sine waves have no harmonics. They're "pure"...just one frequency). So what? Think about it. What is the purpose of a lossy codec? TO SAVE BITS, so that data can be used more efficiently...because there are fewer of them. Well "squaring" a wave makes it more complex, not simpler...which means it's more difficult to encode...not simpler. Arfifacts from lossy codecs are far more likely to sound "fuzzier" than "sharper" or "sparkly". Create a 20kbps mp3 file and see for yourself. There are plenty of artifacts, NONE of them look or sound like square waves ("sharp', or "sparkly").

By the way...when you talk about "squaring" the waves when viewed in an editor, are you talking about zooming in to the sample level? ALL digital audio looks "squared" if you zoom in enough, because digital sampling is in discrete steps. The reason I suspect this is that I just zoomed in until I saw something that looked "squared' on the file. What I saw were individual samples. You know, of course, that with 16 bit audio there are about 64,000 discrete audio levels. Every instant the sound is at one of these discrete levels. There's nothing in between them.

Yea, I understand about overdriving the system, and I see that there is plenty of headroom in the sample, and I know I'm not overdriving anything on my computer. I also know that the sample sounds the same on my iPod, even with headphones at a low volume.

The places where the waveform looks un-naturally clipped, the amplitude isn't really close to the peak.

I'm going to post some more of these pictures (set up a web page), but here are two pics:

First, Mike's voice:
To my ears, this sounds natural, un-compressed and straight off the mic.
It looks like a natural waveform, right?
http://radioguy.googlepages.com/Mikesvoice2.jpg

Next, the voice of the WTQR guy:
This is probably the worst sound of the announcers.
If you notice, that WAVFORM IS NOT NATURAL!!! ...and it don't look like straight out clipping to me.
http://radioguy.googlepages.com/WTQRAnnouncer.jpg

I've got about a dozen other pics like this, where you see one level for a few samples, then it bolts up to another level for a few samples. These are passages that have that weird sparkley sound.

I'm not saying that something couldn't be clipping, but every HD Radio sample I have sound just like this.

Just as I suspected. You've zoomed in to the sample level. The corners of each "square" are SAMPLES. Dude, ALL digital audio looks this way when you zoom in that close on most audio editing programs. Sorry ;)

However, it doesn't playback that way. These 'corners' will be smoothed by the ultrasonic filters in the dac, so actual playback will be smoother than it looks. In most editing programs, a sine wave will look "squared" if you zoom in to the sample level. There IS an exception! Adobe Audition actually takes into account the filtration on playback, so that if you open a sine wave in Adobe Audition, the waveform will look "rounded" between samples...which at the encoding level is quite impossible. There is NO resolution between samples, so anything "between samples" (an oxymoron!) would have to be a straight line. Confused? It's really simpler than I'm making it sound. The "corners" are at a higher frequency than 20khz. Well, the passband 'corner' for 44.1khz digital audio is just a bit beyond 20khz (depending on the filters), so once the waveform is filtered, the "corners" are removed.

Imagine a projected digital image. Get too close and you see the individual pixels. Now go to the projector and defocus just a little bit...until individual pixels blur. Since there is no information BETWEEN the pixels, it's actually possible to smooth the image without losing real resolution. That's a pretty good analogy to what ultrasonic filters do in digital audio. It blurs the "information" (the "straight lines") between the "corners" (samples), until they look "smooth" (or curved..in theory SOUND curved...another oxymoron because the "straight lines and corners" are either too high to be audible, or too small in level to matter). ;)

Ok... I understand what your saying, and have understood that all along.

Basically what I'm saying is that in a natural waveform, you wouldn't have plateaus like that, but rather they would go up from one point and come back down on that same point,
I can't really prove my point without directly comparing the same passage in lossless and HD Radio (or lossless vs. 48 or 96k AAC+ which is what I'm gonna do).

I'm doing side by side comparisons between the lossless Java sample, and the 48k AAC+, and it confirms my suspicions, but the compression used in it does NOT sound the same as what I was hearing in those samples. So... Unless I get my hands on an HD Radio encoder, it's still just an observation.

I'm sorry, but all I see is just the straight line from sample to sample to sample. Even at 48, or 96k you would see the same thing...the samples would just be closer together, so you'd have to zoom in more. Only if you could filter out the ultrasonic component would the waveform start to look "rounded". There was an article about just this subject in a magazine I read a couple of months ago, but I'll be damned if I remember which one. I'll try to find it.

Sadly, the only program I have on my system that can display a waveform is Adobe Audition, which "rounds the corners", as was discussed in this same article (?) Any other audio/digital geeks wanna' chime in here?

Ok... not to keep going round and round about this same subject, but my simple point was that if you notice... a sample of your voice has no flat, straight lines in it. The lines are all slanted. With a little bit of "rounding" from the filter, you voice resembles are "natural" waveform.

If you look at the other waveform, it looks more choppy.

As I said before:

Basically what I'm saying is that in a natural waveform, you wouldn't have plateaus like that, but rather they would go up from one point and come back down on that same point. I also see that the sample before one of these crests would be higher in an un-compressed stream.

Even with filtering and rounding, the waveforms I see on the HD Radio samples seem kind of weird looking. Some of them (not necessary the one I posted) look like they are "stair stepping" in heavy modulated passages (stay at one amplitude for two or three samples, bolt up to another... when the natural waveform would have consistent amplitude increase).

I think there is some additional smoothing and filtering that may be applied by the tuner to compensate for this.

The reason I say this is because I can hear the same kind of sparkle in other HD Radio samples (mostly taken from a Boston Acoustics), but it is not as apparent to me.

This could mean three things: 1. The filtering on the Accurian is different (worse) than that on the Boston Acoustics. 2. You've got a better sound card than the people who did the other samples. 3. All the stations in your area need to work on their signal processing.


As for hearing what I'm talking about, play some HD Radio (with the accurian) over a good sound system, then CRANK the treble. Hear it?

Now Mike, you wouldn't be pulling a prank on these folks like you did the guys on the radio-info.com HD Radio board in the "HDT-1 Review" thread, would you?

[URL=http://www.radio-info.com/smf/index.php/topic,62209.msg440526.html#msg440526[/URL]

By the way, I mean no malice by that.

Also, for those interested, J & R now has the Sangean HDR-1 for $219.00

Tom I'm not quite sure what the "prank" would be. The point on the other site (actually I did it on this one too) was to show that people often claim to hear differences where none exist...as (in my "prank") all three files were identical. Sure enough, some heard "differences"...prefering one to the other two. I did the "prank" to show how fickle hearing and biases can be...they can actually lead us to hear differences which dont' exist.

At any rate, and with respect, all I see on the pics you've shown are straight lines from sample to sample. You seem obsessed with the shape of the lines, but note that lines from sample to sample can ONLY be straight (if shown as they truly exist) because there is NO information "between" samples which would lead to "rounding".

Voice waveforms look different than music. They're far less complex. Perhaps that explains what you're seeing? And yet spoken word OFTEN reveals artifacts (AUDIBLE ones, who cares what a part of a waveform that's 1/44.1 thousandth of a second long LOOKS like! This is the difference between "audiophiles" (whom I have a long loathing for...because I used to be one!) and LISTENERS who just want to enjoy the program. LISTEN FOR DEFECTS AND YOU'LL SURELY HEAR THEM in every recording and/or broadcast. SO WHAT? Is the "window" (the recording) clear enough to see (hear) what's going on? What else matters?

[Let's set aside the whole waveform thing for now]

This is true. I'm don't care whether it's analog or digital, I just want it to sound good. The reason I complain about the HD Radio samples is because it distracts me from the music.

If that's what HD Radio sounds like, I'd prefer if we all go back to analog FM and stop complaining.

I'm more about "acoustic transparency" than anything else. I don't care about pops and clicks, surface noise, tape noise and (low amounts of) static, so long as the audio is decent. I can not stand: really audible [bitrate] compression, bad dynamic range compression and excessive amounts of hiss.

To me, HD Radio isn't really as "acoustically transparent" as I'd like it to be. (as least through the Accurian tuner).

BTW, at least according to Wikipedia, HD Radio DOES NOT use the AAC+ compression algorithm.

[see: http://en.wikipedia.org/wiki/Hd_radio]

"The audio compression algorithm was initially set to be PAC when iBiquity's standard was first approved by the Federal Communications Commission (FCC) in 2002, but the system was changed to the HDC codec in 2003. The change was made because the low-bitrate audio for AM stations was described by some as sounding as if it were "underwater", plus the fact that the partially in-house HDC format has newer patents that can be exploited for longer periods. HD Radio stations must pay royalties each year to iBiquity, plus the costs paid by the manufacturers of the transmitters which are then passed along to the stations that buy them."

First of all, yes HD does use the HDC codec, which if you'll research a little, you'll find is a kissin' cousin to aac+. They're functionally very close to identical.

Second, I guess you expect me to argue with your assertion that analog, with 40db poorer noise performance, 50db poorer separation, and ten to a hundred times more distortion is better. I'd do as well to argue with Jerry Falwell about creation vs. evolution, or that the world is in fact BILLIONS of years old rather than thousands. What's the freakin' point? The Earth is flat, I get it. LPs are better than CDs, analog is better than digital, and higher primates don't resemble us at all. Uh huh. Dude, whatever! If you prefer analog, IT'S YOUR MONEY! I would however point out the name of this board "HD Radio". If you don't like HD radio, well there are other fine boards here, including ones about analog. You'd certainly be welcome in those (as you are here, of course). Forgive me if I take the "analog is better' argument about as seriously as I do creationism.

One point I will counter...about "dynamic range compression". Have you freaking looked at the WDAV audio on your waveform analizer in anything other than the mode where you zoom in to the individual sample level, and are shocked that straight lines connected to other straight lines have sharp corners? The dynamic range is MASSIVE. Listen to the sudden jump in level at about 16 seconds in. Start listening at 10 seconds in, and listen to 20 seconds. If you've heard any dynamic contrasts like THAT on analog radio lately, I'll eat your radio! (I didn't even mention the point at which the announcer beigns speaking, when the recording has faded to near silence, and the level suddenly jumps more than 40db! A FORTY DB DYNAMIC JUMP. Has anyone here EVERY heard anything like that on analog radio?)

What about when the announcer speaks, and you can practically tell that she has a piece of lettuce stuck between her teeth? People who don't work in radio or pro audio don't know what audio straight off a board (not having gone through any processing, and not having been broadcast through analog OR digital gear) sounds like. Well THIS IS WHAT IT SOUNDS LIKE!

Ok, first of all... I'm neutral on the whole analog v. digital debate. Yes, LPs could theoretically sound "better" but I don't think the small extra gain is worth spending zillions top notch playback equipment.

Guess we shouldn't debate creationism then (me being a Charismatic-Christian Conservative and literal 7-day creationist)

That being said, I have a TT and records because it's a super cheap way to get old music, and LPs sound worlds better than tapes.

I love the concept of HD Radio, I just don't like the sound of the compression (er... the way it sounds on an Accurian). All the same, I'm going to purchase the next generation Sangean HD-Radio tuner when it comes out this fall or summer.

As for Creationism, George Will said it best "Christian Conservatives are entitled to their own beliefs, their own methods of worship, their own churches, their own congregations. NOBODY is entitled to their own science."

I know of no "theory" based upon SCIENCE which would make lps audibly superior. They are measurably superior in only one area...frequency response betyond 20khz. And that has lots to do with source material, and how they were mastered. In every other way they are theoretically, measurably, and audibly (when compared to the original source material) inferior to digital, if one equates "superiority" with FIDELITY TO THE ORIGINAL. Well made digital recordings are indistinguishable (in double-blind testing...you know, SCIENCE) from the original source (tape or disc).

If you equate "superiority" with sounding SUBJECTIVELY better, without reference to the original, then yes...turntables may sound better TO YOU. No wonder! They're more "harmonically rich" because of resonances in the stylus/cartridge body/tonearm/plinth/platter/record. They're also more "spacious" because of random phase information introduced by said resonances. Ambient information is more apparent on analog recordings for a logical reason...IT'S AT A HIGHER AVERAGE LEVEL due to the restricted dynamic range of analog technology. Compressing the dynamics of a recording so that it fits within the limitations of the format being used is kind of the job description of the Mastering Engineer. It's a large part of what he/she does. LPs are compressed so that they have less dynamic "swing" than cds. Period.

Creationism - I've got some things to say about the subject, but the whole subject his completely off topic, and I really don't care to debate such because we both have our minds completely made up.

Vinyl - this is a revelation to me (seriously). I've gathered bits an pieces here and there about the whole vinyl v. CD debate, but all the "professional" opinions I've read are written by people who subscribe to the three V's, and believe that putting you $10K speaker wire up on pyrex dishes, using tyce clocks and arranging your audio system according to the laws (er whatever) of feng sui, you'll get a 4 X 10(23)% increase in "spaciousness."

Some more sane folks believe in the superiority of vinyl, but not in that of tubes and the rest of the analog audiophile lore, so this kinda swayed me the other way.

All the opinions I've read to refute this basically site the reasons why analog stinks, and not hard scientific fact to explain the difference.

...and besides that, some of the most stupendously great artists of our day (namely The Flaming Lips and The White Stripes) press on CD and vinyl.

Unlike the debate between origins, there is very little "hard scientific fact" in audiophileism, so it's a bit... confusing.

I'm into vinyl not because it's superior to CDs, but because the stuff I pick up at Savers sound good enough to me, and it's vastly superior to tapes of the same recording.

Hey I didn't mean to get into a big "thang", Matt. I apologize for stirring things up. I believe that science is on the side of digital as providing more ACCURATE sound than dragging a rock through a wiggly little plastic "ditch". But I can't deny that vinyl can sound FANTASTIC. What's amazing isn't just that it works at all, but that it sounds as good as it does. I can tell you fifty things wrong with vinyl. But I can't tell you it sounds bad. it doesn't!

Peace!

Tee hee, it's 2007 and I'm still reading the same old CD vs. Vinyl arguments from twenty years ago. :D

I think that the local engineers task of optimizing the HD signal, can make a whole lotta difference in how an HD or an analog station sounds.

The CBS stations seems to sound better than the CC and the Cumulus stations.

Quality of FM sound, that does not cause ear fatigue has become a non-issue for some corporations....That's sad, but true.... and the FCC.... NOT their concern either.:(

And that's very true !

The davidson station at 96kbs was excellent. The 48kbps had better stereo seperation than analog but the high end was harsh and shrill, maybe more so than FM analog.

>>I guess you expect me to argue with your assertion that analog, with 40db poorer noise performance, 50db poorer separation, and ten to a hundred times more distortion is better.<<

Ah, the memories .......

The only thing I'll miss about analog radio is firing up a big 70's tuner, and flinging the dial all the way up or down the spectrum. I just love the sound it makes :)

There are some situations where yes analog can be better but that's more with a visual sense.

Speedometers for example are usually analog...I haven't seen any digital ones since the 80's...just throwing higher numbers yet isn't as thrilling to some to see a needle go from one side to the other.

I think also for pilots it's just more natural to see analog dials instead of digital.

Oh... I didn't mean that. I meant radio broadcasts and stuff.

I'd bet you good money that analog speedometer in your car is actually digitally controlled.

my first car was a (very used) 1987 Subaru GL-10. It had one of those digital speedometers. Whole dash was digital. I loved it! I wish they still did that!

But then, I'm a geek.... (and no I didn't just spend half an hour looking for a photo of the dashboard) ;-}