Limiting analog audio to 5khz for HD really sucks. It makes the sound worse for today's listeners, in order to make it better for tomorrow's, POTENTIAL listeners. It sucks alright. But must the suckage be massive?

5khz audio CAN sound pretty damn dull. A quick measurement of frequency response shows why (duh!) Fortunately our ears aren't laboratory instruments.

The ear can be fooled into believing there's more hf content than there actually is...through a process Aphex has called "Aural Excitation" for three decades or so. Now there are both hardware boxes (from Aphex and others) and software plugins (I used a freeware one called the RGC Audio HF Stimulator for this demonstration) that allow us to "add" harmonics. Let's not call it what it is...HARMONIC DISTORTION. After all, it's "even order" which sounds very musical to our ears. What I did was crank up "drive" and "mix" to 100 percent, generating a helluva lot of extra harmonics in the little over an octave between 2 and 5khz...restoring the illusion of ample "highs'. By the way, NOTHING has been done to the lows...low frequency response is equally extended on all three files I created for the demo.

Below are the three files. All are 320kbps mp3 files, to keep space down, while still having good enough quality to illustrate my point. One is full bandwidth 20khz from the cd of "Year of the Cat"...the song "On The Border" by Al Stewart. It's DRENCHED in highs, but will a very full bass line as well...which is why I selected it. On the second file I have simply chopped off everything above 5khz (using the bandwidth control in Adobe Audition's mp3 encoder). On the third, I used the harmonics generator program as discussed above, added a hard limiter at -.3db to keep "splat" to a minimum from all the extra hf content, and again chopped off everything above 5khz when saving the file. You may be surprised that the 'brightest" sounding file is the one with the "hf stimulation" (harmonics generator) with bandwidth limited to 5khz. So much for having to roll off bass in order to make things sound "balanced".

On the Border 20khz bandwidth, no processing http://www.theproductionroom.net/border20k.mp3

On the Border 5khz bandwidth, no processing
http://www.theproductionroom.net/border5k.mp3

On the Border 5khz WITH ADDED HARMONICS and hard limiting to prevent overload
http://www.theproductionroom.net/border5kexcite.mp3

"Aural Exciters" anyone? If we're REALLY going to limit highs to 5khz, maybe it's time to buy stock in Aphex!

Analog audio for hybrid AM HD Radio signals is bandlimited to either 5 or 8 kHz. Each station decides. For the past year or so, many non-HD AM radio stations have been limiting their audio bandwidth to 5-6 kHz to reduce skywave interference at night and to increase loudness and signal-to-noise ratio within the narrower passband. For these stations, a switch to HD Radio would either cause no noticeable decrease in analog bandwidth or actually increase it.

Brian

I don't really see any point in AM stations using any more than 5KHZ. Why? Because practically all the AM tuners in our house (a component tuner, HT receiver, stereo receiver, Sangean ATS-909 [among others]) don't go much past 5khz anyway.

Personally, I want better S/N ratio than better frequency response.

Also... frankly, I think the excited 5KHZ sample sounded like crap. I'd much rather listen to the normal 5KHZ sample if I had to choose.

I also hear distortion on the "excited" sample. This was a hasty demonstration done in software. Obviously a well-tuned box from Aphex would work better than an old, free directx plugin from the 'net. My goal wasn't to show you the "ultimate in narrowband audio". It was to show that with "excitation" (or added even order harmonics) a 5khz band-limited signal CAN sound brighter, and more intelligible than an unprocessed one.

Here's the best argument for "excitation". Most AM content these days is talk. The sibilant sounds which our ears use to identify, and clearly discern what's being said are largely at, and above 5khz. Excitation increases sibilant energy in the range around, and just below 5khz...a range that can still be broadcast (on analog) even with HD. It makes talk clearer and more intelligible, even through the very small 5khz "pipe". THAT'S the best reason to use it ("excitation") now. It will make talk radio clearer, and more intelligible.

You say "most" radios have no audio above 5khz? That isn't true. NO if filter (except the most expensive ones) is an absolute brick wall. SOME content passes through the filter all the way to probably 8-10khz (though it may be attenuated 10-20db or even more). This is still audible, and still contributing to perceived "clarity". AM stations CAN (and once were) equalized to flatten response to 8khz on AVERAGE AM radios. It took a lot of eq, and really sounded "sizzly" on broadband radios, but IT WORKED. So yes, the audio on most radios may be WAY DOWN by 5khz...but most radios are still producing SOME audio a lot higher than that.

I think your information about the spectrum of the human voice and how much "bandwidth" we need to discern what is being said is largely anecdotal. My own opinion is that it is down around 2.5 kHz, well within the 5 kHz bandwidth of most radios. We should let the audiologists weigh in here.

While you can't say that there is no audio above 5 kHz in most radios (and I do not mean to imply that you did) the "no" part is the word that gets you in trouble. NPR did some tests as part of the NRSC process and found that the average response of AM radios of varying types was down 3 dB at 2450 Hz and 10 dB down at 4100 Hz. Interpolating their the range from 8 to 10 kHz was about -25 at 8 kHz and off the charts at 10 kHz (lower than - 32 dB.)

Yes, that's still something, but my seat-of-the-pants measurement says that the background noise at the desk in my office is about 47 dBa. When listening to music over my computer's speakers I measure about 57-60 dBa. From those estimates you can say that anything more than 15 dB down (coincidentally about 5 kHz according to the NPR data) is below the ambient noise in my office and largely unheard. Cars are probably worse, as far as signal-to-noise goes.

Yes, stations used to boost the high frequencies of their signals. Most radios are designed with narrow IF sections to allow good sensitivity and selectivity. They are narrow to keep adjacent channels from being heard. The station I worked at boosted their highs to force their way through this filter, making the situation worse for adjacent channel stations. The NRSC tried to negotiate a cease fire in the war, and it seems to have worked. An honest 5 kHz is very listenable for speech and a lot of music, too. Once the AM band got crowded, Hi Fi radio went to FM (at least for about 20 years.)

If you've got a radio that can detect higher frequencies and have a quiet enough band to tune in, great!

Mike.. your right. Radios don't all have brick walls at 5KHZ, but there's not much substance above that mark.

I did a little experiment with a couple of tuners (my Kenwood component tuner and the AM section on my Technics receiver). I tuned to the same station and found that the Kenwood had considerable HF response, although the only real difference was that extended HF response yielded a much high level of interference (our WiFi router wreaks havoc on the entire AM band).

I wonder if the AM stations in your area are already limiting audio to 5khz, Matt. Wouldn't be surprised.

By the way, I didn't mean to imply that voice wasn't "intelligible" with bandwidth as narrow as 2.5khz. Only that it's far more intelligible if bandwidth is a little broader than that. Any production lizard (er...wizard) knows that there's quite a bit of hf content in human voice...octaves beyond what today's AM radio lets us hear.

As for HF response, my benchmark is still the Sony SRF-A100 that I bought in 1983 (it still has the "new car smell" and "It's a Sony" sticker attached"), When they were running measurements on the transmitter at WNNC in Newton NC in '83 after the installation of C-Quam, I had the output of the radio run into the board so I could see the frequency response of the radio. When they ran the 10khz tone, it was pretty much flat...beginning to rolloff only at about 12khz. With the pre-emphasis used on AM in those days, it's response was pretty much the same as FM stereo, and it sounded like it. By the way, I measured actual separation greater than 35db on that radio...when either the left or right channel was modulated alone at 1khz). The station had (still has) a Nautel solid state transmitter...brand spankin' new at the time, CRL processing, and an audio chain that was clean as a whistle. The tower was tweaked and broadbanded. C-Quam left the tower squeaky-clean. It remained that way for probably 30 feet at night ;)

Woah! response up to 12K! If all AM stations sounded like that, I'd have different opinions on AM radio.

BTW... I did a little bit more complete comparison of tuners (Sangean ATS-909, VS. my Technics receiver vs. Kenwood tuner), by running an A/B/C test, and found that there was a significant difference between the three - the kenwood having much better HF response (which made speech more intelligible).

Woah! response up to 12K! If all AM stations sounded like that, I'd have different opinions on AM radio.

BTW... I did a little bit more complete comparison of tuners (Sangean ATS-909, VS. my Technics receiver vs. Kenwood tuner), by running an A/B/C test, and found that there was a significant difference between the three - the kenwood having much better HF response (which made speech more intelligible).


Problem is (and was): most AM radios still sounded like junk because they cut off way too early. AM stereo on narrow band radios sounds as bad as mono AM on narrowband radios ... only in stereo. The infamous Radio Shack Realistic TM152 comes to mind (I think that's the number). Looked nice. Sounded awful. No wonder AM stereo never caught on.