x
dpd 16 February 2006

Hi guys & gals...

I've been toiling a bit at the radio station for the last couple weeks limping along with our Classical station audio after our 20+ year old Orban audio processor puked one day.

I first tried to get buy with nothing, but just couldn't balance maintaining the maximum deviation limit (105%) and the noise floor. The sound was either clipped or you just couldn't hear it on the air during the quiet sections.

Next up was to use the same Orban PC1100 PCI card I use for our Classical webstream. I got things dialed in OK, but it's still half-assed. Clean, but still needs more compression to get over the very low power we are allowed, so I'm not getting much boost from the processor. So, I recommended a new FM processor which is ordered.

So, while I was lamenting why it's so damn difficult to compete on the dial with the big commercial FMs - telling the GM how/why these guys sound so LOUD and why that would destroy the Classical format if we copied them - I got an idea for an experiment that I conducted earlier tonight.

I took the output of my tuner (a Boston Acoustics Receptor HD unit) and piped the input into my Pro Tools rig at home. I recorded 60 seconds of music on 20 FM stations on the dial - without adjusting any levels on the record side.

Wowsers! The waveform views confirm what my ears have been telling me, but far worse. Nearly every one of the waveforms (except my three signals - NPR Talk, Classical, and a satellite Jazz service) are SQUARE WAVES. Crammed up against the limiter the entire time and running peak FM deviation., be it rock, oldies, christian, country, or urban formats. I have to do a bit of analysis, but I can pretty well guarantee that at least two stations are greatly exceeding the FCC maximum deviation. (I'm debating about borrowing a HP modulation analyzer from work to confirm this.)

So, all you Mastering and Mix Engineers are just having your work ameliorated by this overcooked processing. Imagine seeing a waveform from a 60's tune coming out like a square wave. It doesn't matter what you do before it hits the air - you have ZERO control over how it sounds when it airs. ZERO.
mode>

So, what I really want to know is - what do you guys want to see from this experiment? The waveforms? Spectragrams? RMS levels? It may take me a few days to get some graphics of this, but give me some ideas and I'll try to show you reality.

Topic Tags

Comments

RemyRAD Thu, 02/16/2006 - 22:55

Not only do FM stations use multiband or spectral processing, they also use "composite clipping" to squeeze an extra couple of DB's for more loudness to compete in the loudness wars of commercial station ratings! It's a trade-off between distortion and loudness. Stations are allowed a low percentage of strict over modulation. Composite clipping does not have anything really to do with audio but rather with the baseband composite FM signal but of course also affects the amount of distortion within the broadcast waveform. LESS IS MORE! Public radio stations and especially the classical ones DO NOT NEED TO COMPETE FOR RATINGS IN THE COMMERCIAL MARKETPLACE! What were you thinking??? If you want to play loudness wars go get a job at a rock-and-roll station that plays commercials. If you want people to enjoy their classical music fine arts listening experience, use some gentle multi-band spectral processing and don't push your luck or your signal, you will only lose your listeners. I hope you're not the chief engineer with your above perceived misconceptions of good sound?? I won't tell your chief engineer if you go and do your homework quietly.

1st phone FCC Mother engineer
Ms. Remy Ann David

dpd Fri, 02/17/2006 - 19:19

Michael: Working on it tonight. Plugged in InspectorXL on each station's track to get some spectral data and RMS/Peak comparisons. I'll need to do some massive Windows capture that I'm not yet set up to do, but I'll dig around and get something going. Played back a lot of it and just sat there and shook my head on some, if not most, of this crap. The worst offender had about 6-7 dB between 'average RMS' and peaks.

Remy: I'm not the CE - I'm a real Engineer. I do this gig for mad money and fun, but I'm not laboring under false pretenses here. I've been observing the FM loudness wars in this market for ~20 years, but this is the first time I have ever measured anything like this. Yeah, we don't compete for ratings, but getting the loudness optimized for our formats is still important and we aren't optimized, based on my measurements and listening.

This IS my homework - I'm attempting to see how much peak and RMS headroom I can use up without trashing my sound on the top end so I can overcome the puny little signal my 43 watt (yes, you read that correctly) FM classical puts out from what I believe is a pretty low-end exciter the CE ordered for this station. I'm not convinced that the modulator and/or stereo generator in this thing has a low enough noise floor. Maybe it's a setup issue, maybe it's a design issue.

I really don't think I want to go with multi-band compression on the classical. I've found that two-band (at least on my old Orban) sounds better. I'll find out how the new Omnia 3 fmt (only $3K) will sound sometime next week once I get it installed and start tweaking it.

ps - Remy? You aren't too good at Hateball, are you? :wink:

dpd Sat, 02/18/2006 - 21:51

Brandon wrote: here in chicago just about every major station is "now broadcasting in crystal clear hd radio"

http://chicagobusiness.com/cgi-bin/news.pl?id=17515

interesting technology

have not heard it yet myself, but i bet the compression wars will sadly ruin this format also.

I kept warning the GM but he insists on saying the following on our spots:

"CD-quality sound, High Definition Radio..." Gag me. We are the 1st FM station in Indiana to multicast in HD. Here's how it breaks down:

- HD1 - Main Program Service (64 kbps) - NPR News/Talk/Jazz
- HD2 - Supplemental Program Service 1 (32 kbps) - Classical
- HD3 - Supplemental Program Service 2 (20 kbps) - Jazz service

At times we transmit the same programming on Analog FM, HD1, and HD3. I have both analog FM and HD tuners at home. The difference is scary. Calling HD3 anything but 'compressed, lossy' is false advertising. Our web streams are only 22 kbps and they sound about the same as HD3.

The extremely talented, hard-working, people in the recording industry - AND the artists - deserve better. When Analog FM goes away, the bandwidth will be available for IBOC. Some may opt for higher bitrate codecs, but some will be tempted to add even more, lossy, reduced bitrate streams and forget sound quality.

Of course, these streams will still be compressed, clipped, limited, and phase-rotated to death.

Brandon Sat, 02/18/2006 - 22:20

wow!
that's terrible!
is the format even capable of 16-bit/44.1k stereo pcm?

dpd i am a software engineer
i have been working on a 24-bit cd for about a year and a half now.

it does work, but now that 3 bands are almost ready to go forward with this, i have changed the entire thing.

we will be releasing these as a cd/dvd combo

most likely in DualDisc format.
it's an audio dvd, not dvd-audio, confused yet? lol
24-bit/96k stereo pcm, plays in any dvd player.

anyway i was hoping you might have some insight for me.

is this something you would be interested in helping me out with?

can call me anytime 630-580-4115
or email twentyfourbits@yahoo.com

ok guys sorry for getting off topic -- back to FM bashing lol

dpd Tue, 02/21/2006 - 18:38

Tabulated Data

OK - I pulled this data off InspectorXL. I apologize for the format - not sure how to post a proper table.

The data across all stations was normalized to -1 dB FS Peak. I'm estimating that -2.5 dB peak is legal FM deviation, so some of these stations are clearly above the limit! (Up to 20% over the limit, if my calculations are correct)

Freq (MHz) Format Peak RMS Peak RMS Content
88.3 Christian -2.3 -13 -10 pop rock
88.7 Classical -9 -24 -20 baroque
89.1 HD1 Talk -5 -21 -18 NPR ATC
89.1 HD2 Classical -15 -30 -24 baroque
89.1 HD3 Jazz -6 -20 -17 straight ahead
90.3 Christian -1.6 -13 -11 pop rock
93.3 Country -2.6 -14 -10 Reba McIntire
95.1 Rock -3 -13 -11 generic rock
96.3 Urban -2.2 -12 -10 synth hip hop
98.9 Rock -3.2 -13 -10 acoustic rock
101.7 Oldies -3.2 -15 -12 Johnny Rivers
102.3 Rock -1 -10 -8 grunge rock
103.9 Urban -3 -15 -10 hip hop
105.1 Country -3.4 -15 -12 country rock
106.3 Oldies -1.5 -14 -11 60's oldie

The peak is direct from the meter, the RMS and Peak RMS are eyeball readings over a 10-15 second section of the 1 minute period. (The "Peak RMS" is just the max RMS in that time period.)

My stations (88.7 and 89.1) are sucking wind. 88.7 due to use of our web stream processor while I await delivery of a new Omnia 3fmt. 89.1 due to not fully understanding how to best process for HD Radio. The CE doesn't seem to care as long as they are on the air. sad

Many of these stations are banging off the limiter so much that they sound really ratty, to boot. Loud? For certain, but I was cringing listening to some of these - if only for a minute.

x

User login