Frank Foti, David Bialik
This joint meeting with the New York Chapter of the Society of Broadcast Engineers took place in
the Radio City Room at the Microsoft Conference Center. Mr. Foti covered various aspects of the
FM Stereo transmission chain. He began by discussing the necessity of putting in more than 10,000 hours of
time when brainstorming new concepts, as posited by author Malcolm Gladwell. Such is the price of innovation.
The elements of great audio broadcast processing include pleasing level control, foolproof level detection,
rock-solid equalization and tonal balance, dynamics with low intermodulation distortion ("IMD"), and
quality competitive limiting. Properly controlled, these factors will provided listeners with an effortless "loudness experience."
Getting levels right "all the time" involves minimizing IMD. Limiters are peak responding and those for loudness yield IMD. The human ear is an RMS (averaging) detector, and perceived loudness is generated in the AGC (automatic gain control) and is governed by compression. Old-school RMS using a single averaging coefficient is inconsistent. Frank's new method uses self-adapting RMS which examines the content of accelerometer/decelerometer coefficients, ignoring the short-term peaks and resulting in consistent RMS. Less limiting is required when the AGC action is correctly determined. Deep limiting of 15-20 dB is a thing of the past. More detail is delivered to the listener because the AGC and limiter are delivering consistency to the clipper circuit.
Another area under development deals with undoing the fatiguing effects of the hypercompressed audio found in many contemporary music productions. Declipping of these square waves removes interband coupling. Frank discussed the concept that dividing the frequency spectrum for processing purposes into more than six or seven bands yields diminishing audio quality returns. Factors which may overtax limiters include crossover phase errors, incorrect time alignment, and recombination/summation errors.
The 75µsec pre-emphasis curve, which boosts 15 kHz by 17 dB, must be included in testing procedures. A static approach to limiting is not sufficient and the result is clipper-induced IM and harmonic distortion. Autosensing is necessary to address the linkage between the limiter and the clipping system. Controlling the duty cycle with the pre-emphasis and not the peak level keeps the sound smooth.
Another method involves superior stereo pilot protection. Multipath distortion can be audibly reduced by relocating the L-R pilot from 53 kHz -38 kHz to the area between 23-38 kHz vs. The cleaner multiplex spectrum thus achieved will protect RDS, SCA and HD-1 signals.
Frank presented before-and-after audio demonstrations, along with scope pictures, of the concepts discussed tonight. Such informal listening tests have begun at regulatory agencies. Applying these innovations will defeat the blend circuits in conventional receivers, thus opening up the frequency range and improving stereo imaging. The multipath effect is also greatly reduced. The technology includes transitioning from dual sideband to single sideband stereo transmission. Frank cautions that there may be as-yet unknown gremlins which will appear in future tests, but these will be tamed as well. One dB more gain may be obtained by embedding the stereo pilot in the main channel. As the algorithm occurs only during final limiting there is no affect to the pilot amplitude or spectrum, thus yielding the extra dB of loudness "on the air."
A lively Q&A period followed the presentation.
