The author is U.S. sales manager-radio processing sales for The Telos Alliance.
I gave a fun talk for a bunch of engineers in Los Angeles recently and wanted to turn it into a written piece.
There are a number of common factors that tend to diminish audio quality in radio facilities. It’s just my opinion, but the biggest “quality robbers” at radio are:
• Too many A/D and D/A conversions in an air chain
• Lossy compression in the studio-to-transmitter link (STL)
• Lossy compression in source material at the studio
• Pre-clipped source material
• Using an exciter to apply stereo coding or final limiting/clipping
• Lack of normalization strategy in the automation system
• Not knowing your total modulation, or the competition’s
• Other processing units in the chain set for “effect”
You should avoid multiple D/A and A/D conversions whenever possible. View it as “generation-down,” just like we did in the analog tape days. And whenever possible, convert once (as best you can), and keep it digital. It’s one of the many benefits of using Livewire … cleaner more widely-available audio … instantaneously.
“LOSSY COMPRESSION” IN YOUR SIGNAL PATH TO THE TRANSMITTER
It doesn’t matter which codec (or rate) your bit-reduced STL system might use. Unless it’s linear, or you are already running uMpx, my apologies. There are far cleaner linear alternatives available today, including rolling-your-own IP-based radio network, or even our own revolutionary new product: uMpx.
Forget everything you’ve learned about bit-reduction: With uMpx, you can keep your processing at the studio, and send your entire signal, including RDS or SCAs, over a 320 kbps IP bitstream … with no overshoots and no common compression artifacts. That’s a bold statement. uMpx support is built into our software-based Omnia.SST processor and in an upcoming version of our popular Omnia.9.
“LOSSY COMPRESSION” AT THE STUDIO
Granted, not all codecs or bitrates are created equal. But none of them beat the real thing when it comes to the source material at the studio side.
My best advice: Avoid all use of MP3, MP2, etc. … unless emailing a spot to a client. Teach the staff that unless the file size is basically 10 megs per minute of audio, something is not right! I’ve seen record labels send over WAV files of new songs that were clearly once MP3s.
Programmers: “What if I want to clean up or replace my entire music library?”
CLIPPED SOURCE MATERIAL
Today’s popular music has reached new levels of brickwall limiting and clipping. We can’t go back in time, so hats off to Omnia’s Hans Van Zutphen for creating the declipper/undo capability available in our Omnia.11, Omnia.9, Omnia.7 and Omnia.SST software. It absolutely helps. It can’t fix everything, but it will absolutely add back some missing punch and excitement to the overall sound.
USING THE TRANSMITTER OR EXCITER TO HANDLE FINAL LIMITING/CLIPPING AND STEREO PILOT
There may be a multitude of reasons why your plant is set up this way, but let the processing handle the processing. Truthfully, it makes me a little sad inside when I see the stereo generator and final clipper of one of our processors bypassed, only to have these vital functions handled by some transmitter. Use ours, instead. An Omnia.9SG works great as a composite clipper and stereo generator if you want to keep another processor at the studio.
Level management didn’t go away when old-style VU meters started disappearing, and everything went digital. With modern automation, the best we can do is educate our people moving forward and try to cohesively import or reimport audio at normalized, strategic levels.
In short: normalize your music, at one standard level (–10 or –15), your imaging and spots at a slightly higher level, and perhaps your voice tracks or on-air mics at a slightly higher level. You’ll achieve a smoother sound and “surprise” the processor a lot less, especially if you can normalize to dBRMS, which takes into account a file’s overall loudness and peak level.
It’s always best to know what you’re up against in your market. Nothing beats a frequency agile, well-calibrated modulation monitor. You’ve seen the station that ran 120–130 percent total MPX modulation, or worse, right? Receiver designs have changed. If loudness is your goal, it pays to be legal more than ever.
I recently helped a group in a pretty good-sized market figure out that the rest of their market wasn’t overmodulating after all. It turns out their mis-calibrated mod monitor led them to only achieve 90 percent total modulation on all four of their main FM 100 kW signals.
I saw a station in a top-10 market recently where the stereo pilot injection reading was flying continuously between 1–10 percent. The pilot was turned on at the transmitter and in the processor! You just never know until you take a good look at all of it!
OTHER OUTBOARD PROCESSING BOXES
If you employ something like an Aphex Compellor before your STL, and it’s set to compress instead of level, dial it back. And lastly, without naming any names or pointing any fingers, if you turn something else in your chain up to an extreme level, there is a price to pay somewhere else.
OTHER QUICK PROCESSING SUGGESTIONS
How does dry voice sound? Consider backing out of your mic processors.
Don’t adjust too much at once. Save presets frequently, so if you get out in the weeds, there’s a way back.
Stop tweaking after 35–45 minutes. Ears need a break, I know from personal experience. Go for a walk. Do something else.
Check modulation regularly along the tweaking/adjusting process. Listen like a normal human being on a variety of radios. Does it sound noticeably different in mono?
Programming and engineers: develop an open dialog about your sound.
Avoid tuning your processing to make “that one song” sound good.
And did I mention, stop using MP3s already?
Thanks for reading, if you have any other tips or suggestions, I’d love to hear from you.