A Seminar presented by Greg Burnett
at Audio Fest 2004, Colorado Audio Society
by John Haralson
October 9, 2004
Why are audiophiles interested in FM radio? Because the quality of FM audio was satisfying until stations began to compete with each other to maximize the loudness of their broadcasts – the "modulation wars", as Burnett calls the phenomenon. Burnett's argument is that, through injudicious use of audio compression, stations have "squashed" emotion from their broadcast music. He says the problem is particularly serious in Denver and Colorado Springs, where most stations process their audio heavily. Burnett observed that, as recently as 1997, FM audio standards in Los Angeles were much higher than in Denver. He attributes the heavy processing in markets such as Denver to "inbreeding and, in some cases, incompetence" amongst general managers, program directors and engineers: radio stations insist upon sounding as loud as they can, and they are willing to sacrifice fidelity and dynamic range to achieve that goal through compression.
To explain compression, Burnett began with a description of how FM radio works:
In the U.S., ±75 kHz deviation is the FCC standard for FM broadcasts. An assigned FM channel is 200 kHz wide, so the 75 kHz deviation on either side of the center of the channel occupies 150 kHz of the 200 kHz bandwidth. Other program sources, such as Subsidiary Communications Authority (SCA) and/or IBOC ("HD Radio), can extend modulation beyond the ± 75 kHz limits.
Another FCC-imposed standard is a 75 µs (microsecond) pre-emphasis curve. The purpose of the pre-emphasis during broadcast is to compensate for high frequency (HF) roll-off by the de-emphasis network in the user's radio receiver – de-emphasis intended to reduce noise and hiss at mid to high audio frequencies, thereby improving the signal-to-noise ratio (S/N) at frequencies to which the human ear is most sensitive. But Burnett says the pre-emphasis curve used in the U.S. (75 µs) is too steep – too aggressive – and therefore makes broadcasts susceptible to loss of high frequencies when modulation is compressed.
According to Burnett, in an ideal FM program, human voices register at perhaps 40 or 50% on a modulation meter. That leaves "headroom" (up to the 100% level) for sibilance and other high frequency information, which lends presence and emotion to music. Ample headroom also accommodates a wide, satisfying dynamic range.
However, today's FM stations, driven by pressure to sound louder than their competitors, compress their audio through processing, so that modulation is sustained nearly continuously at 100%. The result is degradation of music, as there is no headroom for high frequencies, and dynamic range is constricted.
In order to comply with FCC regulations, radio stations must process their audio by limiting high modulation levels. And they need to boost extremely low-level program material so that it will not be inaudible to listeners. But how much compression of these extremes is acceptable? To what extent should a station sacrifice the integrity of its program material in order to sound louder than its competitors?
Burnett acknowledges that the ideal compromise is a "tricky window". He demonstrated the extremes of no compression and heavy compression with a Orban 8000 processor, through which he fed the output of a compact disc player into a low-powered FM transmitter. A closed-circuit TV camera enabled the audience to monitor the processor's modulation meter. With no compression, and a wide range of modulation, Burnett described the music as "emotional, natural (with 'air'), addictive and wonderful." With heavy compression, and modulation constantly at 100%, the same musical source lacked emotion; it sounded "squashed, stressed, rolled-off, heavy and bulky." (The rolled-off sound he attributed to HF limiting, made worse by the pre-emphasis curve.)
Early processors consisted solely of 75 µs pre-emphasis and a wide-band automatic gain control (AGC) with a slow time constant, which reacted to dynamic changes gradually rather than instantly. But they did not produce the high modulation densities that today's broadcasters demand. Newer processors (from the 1970s and later) have added, as a minimum, a high frequency limiter and a clipper. Digital processors, now available, can "look forward" by delaying the program slightly, adjusting peaks even before they occur. Many other processing options are available, including "multi-band," which divides the audio spectrum into several bands and then compresses each band individually, but with adjustable correlation between the bands. These and other options give stations the ability to tailor their sound. The problem, according to Burnett, occurs when broadcasters "abuse" the adjustments of these options too far towards loudness at the expense of audio quality, failing to meet even the minimum expectations of their listeners.
Burnett acknowledges that a certain amount of compression in an FM broadcast is both necessary and desirable. But it is possible, he believes, for a processor to increase loudness without losing essential dynamic range and high frequencies; compromise must be accomplished very judiciously in order to satisfy the expectations of audiophiles – and even non-audiophiles. Most stations, he says, fail to get it right.
"I wish I could score everything for horns." - Richard Wagner. "Our writing equipment takes part in the forming of our thoughts." - Friedrich Nietzsche