Processing History by J Somich 5

The "Prophetic" Processing History
Yesterday, Today and "Beyond"

Back

Go To Part #6

4/05/09
This series of articles re-published from www.radio-guide.com
issues of February to August 2007 original on PDF format.

Understanding audio processing is a combination of electronics, history, and art. Creating a station’s “sound” brings these concepts together; digital signal processing adds yet more options and flexibility. As Jim Somich writes, our ability to comprehend the totality of what is going on is essential to producing good audio that attracts listeners.
(Radio-Guide June 2007)

Processing
Yesterday, Today, and Tomorrow
by Jim Somich

Part #5

On the Edge of the Future
In the Good Ol’ Days, it was relatively simple to physically modify a processor to change the sound in various ways. Hundreds of engineers had resistors and capacitors tacked on inside their Audimaxes and DAPs. Each of these engineers felt they achieved something that the designer overlooked.

A DIFFERENT PLAYING FIELD
On the other hand, a DSP box is a computer and therefore it is not very simple to make major physical changes in the components. Furthermore, as things stand today, the end user has no access to the software, except for the adjustments or presets permitted by the manufacturer. (This is an issue we will address more directly in a bit.) And there is one more major change evident. Unlike in the past, the sound of today’s stations is often created by non-technical types: consultants and program directors. The engineer may have some input, but he is no longer the decider when it comes to the final sound of the station.

PROCESSING BY PDS AND CONSULTANTS
This is not necessarily bad if the people making the decisions have a great set of ears and know what they want. Unfortunately, too often they have neither. It is then the job of the engineer to educate – and this can result in some contentious confrontations. Some consultants just want to duplicate the sound of a station that “worked for them.” Sadly, many PDs get involved in “pi**ing contests” that help no one, by destroying the sound of their station and forcing the competition to do the same. (I actually have worked with programmers who purposely destroyed the sound of their station, thereby “forcing” the competition to do the same or get lost in the noise. That is no way to run a radio station!) Rabid button-pushing is no way to compare your station with the competition. That went out with manually tuned radios. And, unless the very same CD is being played, it is usually a pointless comparison anyway.

ONE STEP FORWARD, ONE STEP BACKWARD
Years ago Bob Orban remarked that the entertainment industry was the only business he knew of that purposely degraded their product. Since that time, Orban notes that the music industry, in particular, “has started to degrade CD sound to serve the same senseless loudness-fetish that has haunted broadcast signal processing for decades.” I think he has a point. Heavy-handed processing is not the same as turning up the volume. It fact, it is far worse and can, at best, severely limit your TSLs or, at worst, scare listeners away forever. This is probably a good time to remind you that almost any audio processor can be made to sound quite good. Most designers have great ears and they usually develop processing that will please critical listeners when set up properly. But the designers have no defences against reckless operation or setup.

A TWO-PROCESSOR WORLD
At the risk of seeming unfair to dozens of manufacturers, it is quite evident that in 2007 we live in a two processor world. Now, before you call me onto the carpet for making what on the surface seems like a sweeping generality, let me quickly qualify it a little bit: when it comes to top-of the-line stations, in markets large, medium, or small, it appears that most are using DSP processors of the Orban or Omnia flavour. I think I can sum it up by stating that these companies make the best processors and they become the product of choice when cost is not a factor. The fact is the Orban Optimod 8500 and the Omnia-6 represent the leading edge of the state-of the-art in 2007, yet they are two distinctly different flavors of processing. Depending on the processing strategy of your station, most often one will suit you more than the other. Of course, you can fault any product, and certainly no processor is perfect, but these state-of-the-art DSP processors are the pinnacle of processor performance in the early 21st century. At the same time, there are dozens of other manufacturers Worldwide, all striving to enter this processor Winners’ Circle. Personally, I wish them luck. The processing business can always use new ideas. To these folks I say, “Build a better box and join the winners’ club!”

These processors set the pace for the audio processing industry.

BEYOND THE “SAFE CHOICE”
Some time ago, a computer geek immortalized himself by stating that “no one ever got fired for buying IBM.” Perhaps that sentiment might be a little dated today, but you get the idea. The top end of the processor business is a difficult club to break into. Buy an Optimod or an Omnia and your meager little pay check is assured; you can keep your thankless job for a little while longer. Sure, there are adventuresome souls who will put it all on the line for a “new guy” in the processing biz. But if you take a chance on Brand Z and the PD cannot get the sound he wants, you can quickly become the loneliest guy in the station. (You might even be more welcome flipping burgers at Mickey D’s.) The reality is that most of us like to eat regularly and put a roof over the family’s head. None of this should be deemed a discouragement by the other manufacturers and designers. If new ideas were not useful, we would still be using those Sta-Levels and Max Brothers’ units. Bob and Frank did well. And others are doing good work today. The proliferation of processors, especially those aimed at specific users – and their needs – is gratifying to see. So, to all you Corny Goulds, Scott Inczs, and John Burnills out there: I hope you keep translating your aural dreams into new products. Someday, someone, will “beat the pants” off an established processor and – Shazam! – You will enter this select circle of creative engineers.

The challengers to the crown.

PROCESSING FOR HD RADIO
HD Radio is here – warts and all. A year ago, we were discussing processing for HD stations as “future processing” but, from the poor sound of many HD stations, the time to discuss it is right now. In highly data-compressed, coded audio systems, processing becomes even more important; it can make or break the digital sound of your station. The worst offenders are those who use the same processing for both their analog and digital signals. One of the great advantages of HD is the lack of pre-emphasis – there is no need for heavy and fatiguing high-frequency limiting and clipping with HD. Offending stations are easy to spot: there is virtually no difference between the sound of their analog stations and digital streams. Part of this comes from a distinct lack of attention to the digital channel, as it holds so few listeners. Yet, if HD is to capture the imagination of the listening public, many stations must start taking it more seriously. Dead air has to stop. Multiple audio sources running on top of each other have to stop.

TAKING PROPER CARE OF CODED AUDIO
It is also vitally important to avoid any form of clipping when processing coded audio. Look-ahead limiting is the answer – and it works. The processors of tomorrow will address the coded audio problem in very sophisticated ways. (Check out Frank Foti’s thoughts on page 24.) Lastly, there is no need to over-process the HD streams. Now is the time to establish a few dBs of “breathing room.” We are at the dawn of digital radio and much will change in the years to come. But to insure the best possible results with today’s gear, take your HD signal seriously – even if only a few people are listening right now. Think of this as a “dress rehearsal” for what is to come. As an example of the variety of options available, many stations are using PCs with DSP cards like the Orban Optimod PC-1100. Reports from the field are that the results can be equal to using the finest dedicated hardware-platform processor. And we have not really begun exploiting the potentials for broadcasting surround-sound audio.

PROCESSING VISION AND DESIGN
With few exceptions, processors are the vision of one person. Even when multiple persons are involved, the overall focus normally has to be that of one “lead” designer. The reason why processors designed by committee are seldom very successful is that each designer develops a personal strategy. Most of the time, this is reflected by the type of station that buys the product. The closer the processing strategy of your new box reflects your own strategy, the happier you will be with it. The current Optimods are somewhat unique in this aspect: they represent a successful collaboration between Bob Orban and Greg Ogonowski. While Orban’s original vision started the product line and had guided much of its growth, he credits Ogonowski with a critical part in the process. Orban says: “I think this is very important: when there are two people involved in a design and each one respects the other’s abilities, then each designer serves as a sanity check on the other and the combination brings a wider gamut of viewpoints into the design. This is particularly true with digital processors, where there may be dozens of factory presets. It is best when these are formed by the experience and taste of more than one designer.” Ogonowski's input is quite evident in the Optimod, as well as his design work on other recent Orban products, especially the Optimod PC and theOpticodec PC. These have led to some very interesting and innovative ways of bringing higher quality audio to Internet streaming using world-class standards based MPEG-4 AAC/aacPlus audio codecs (the family of audio codecs that helped make iTunes the leading source of downloadable music).

PROCESSING STRATEGIES
As we noted at the start of this article, in the past (before DSP processors) a reasonably proficient engineer could alter the processing strategy of any box to more closely suit his needs. This was often accomplished by changing the values of certain components or by making fixed-value components variable. On the other hand, in a DSP box, you are much more locked into the processing philosophy (strategy) of the designer. There appear to be two rather broad and somewhat overlapping strategies employed in modern day processors: consistency and aggressiveness. These strategies tend to be mutually exclusive. The topology of a processor is usually a dead giveaway to the processing strategy in use. For example, a processor that relies more on clipping tends to be more aggressive and have a more obvious sonic signature. The same is true of a processor that uses more bands. The opposite is also true. A processor relying more on high speed limiting than clipping and has fewer bands tends to be more consistent at the expense of aggressiveness. A sort of middle ground is emerging right now, made possible by the software: a single unit that can select between, for example, two or five band processing. Similarly, with the more powerful computer processors (CPUs) available today, it is easier to implement look-ahead-type limiting and balance it with clipping for better distortion control.

CHOOSING THE RIGHT APPROACH
All in all, the happiest customer is the one who has (wisely) selected a processor with a strategy close to his own. Processor impressions are very personal and I am convinced that we all hear things a little differently. At the same time, all too often the choice of a processor (strategy) and the presets employed (tactics) are a committee decision. In other cases, sometimes only one person – usually the PD – has veto power over anyone else. Of course, you can employ a processor with an aggressive strategy on a format that begs for consistency and pureness, but you will be fighting an uphill battle. There are other ways that the strategy of a processor can be morphed; the most common is to add additional processing either before or after. This was done with analog processing as well, but the procedure acquires more importance in a DSP world where all of a processor’s strategy is etched in silicon. So it would appear that this aspect is where we find the shortcomings cited by most as they work with current DSP processors. Unless you are smart enough (or lucky enough) to select a processor whose designer has a philosophy and a strategy like your own, you will always be a little unhappy with it. Meanwhile, there are a quite a few engineers and programmers with new and different processing philosophies, each of whom would like a stab at trying something a little different. Next time we will don our robes, get the crystal ball out of the closet, and prognosticate about the future.

Go To Part #6

www.261.gr