Commentary: Foti on Preprocessing Audio for HD Radio

This is one of several articles that appear in the Sept. 24 edition of Radio World focusing on studio considerations related to the implementation of HD Radio.
Publish date:
Social count:

This is one of several articles that appear in the Sept. 24 edition of Radio World focusing on studio considerations related to the implementation of HD Radio.

We live in exciting times.

Think about it. Communications and computing technology has changed the broadcast industry in so many ways. The PC and associated networks have dramatically affected how radio programs are produced and distributed.

Consider what we've observed in our industry over the last five years alone: DSB, HDTV and now HD Radio. The last time we've had innovation with similar significance was nearly a half-century ago when color television and FM stereo were introduced.

We have witnessed a lot of change in very little time. We're the lucky ones who get to have the experience of working with these exciting new technologies.

Over the past few years, as development and testing progressed regarding HD Radio, audio processing was one of the key topics of discussion. It became readily apparent that dynamics processing would figure in both the aural and technical performance aspects of the system.

Processing considerations are broader than just the audio chain creating the on-air signature. The HD Radio signal employs data reduction for transmission purposes. Thus, there are issues that must be dealt with at the content creation stage, as well as in the production studio. This is where our discussion begins.

Tools for the task

At Omnia Audio we are in a pretty good position to bring understanding to these topics regarding processing and production. We've been involved with audio processing for more than 15 years, and our transistor-sister, Telos ,is one of the world's leading audio codec developers, as well as the company that introduced MP3 technology to the United Stated.

We've spent considerable time learning, understanding and devising how these two technical functions fit together. Before delving deeper, let's look at a few obvious issues.

First, all transmission media are not the same. Okay, this is obvious, but why do so many people think that using an old FM limiter will suffice to process narrow-band Internet audio? Or why do highly produced spots that sound kick -- ass on conventional FM sound gritty and swishy -- swirly in the bit-reduced environment?

This is what happens when using existing processing and production methods in the bit-reduced world, and it's not the same. It must be understood that coding artifacts are enhanced rather than suppressed when sticking with convention. This holds true, all the more, for HD Radio. (Speaking of convention, we'll use the term HD Radio or HD, when referring to the new digital IBOC signal, and the phrase conventional FM with regards to what has been known as the FM analog signal.)

Audio processing can be a great tool in the broadcast environment. Sadly, it can be misused, too. To put it in simple terms, when it's abused on conventional FM and AM signals, distortion occurs. In the HD Radio realm, misuse will generate exaggerated coding artifacts. For more in-depth discussion on this aspect, point your browser to the following for more detail: . A small segment of that paper appears here.

HD Radio must use fairly aggressive bit-reduction in order to maintain proper bandwidth over the air. At the time of this writing, the FM system uses 96 kbps and 32 kbps for AM. So the challenge of creating great sounding on-air audio is presented at the very end of the transmission pipe.

By the same token, this is the easiest point in the system to break the audio as well. Therefore, anything done earlier in the entire process can be magnified at this point.

Audio-chain processors are designed specifically for HD Radio. They have dedicated algorithms that manage the audio spectrum so that maximum quality is maintained, along with improving codec performance at this aggressive rate. Simply stated, the HD-R path is processed differently from the conventional path as they both require specialized algorithms that are designed to support the tech properly.

Equally important, production techniques and procedures must be modified as well for the same reason. The changes in radio production for HD Radio are not that significant, but the results are. These involve how dynamics processing is used, content quality and the use of prior bit-reduced (transcoded) sources.

Toys in the rack

Let's break each of these down in greater detail:

Dynamics processing: This covers compression/limiting, EQ, stereo sound field enhancement and effects - basically all the toys in the rack, be it physical or PC-based.

Conventional broadcasting for FM and AM is basically a linear process. Within reason, what goes in is usually what comes out. Adding compression and EQ to "phatten up" a station promo or imaging element will print on the air in the manner it was intended. Using the same method in HD Radio may produce different results, and not always for the better. Adding density and EQ can - and more times than not, will - upset the operation of the encoding process in the HD Radio signal. This will yield coding artifacts that may appear as that annoying swishy-swirly sound and/or as added distortion.

Care should be taken to avoid using heavy dynamics processing such as high ratio limiting, clipping of any sort, as well as radical EQ curves when creating a signature sound. All of these functions will generate added coding artifacts. The worst offender is a clipper, as the harmonics generated by the clipping process can drive a codec crazy.

Stereo enhancement is another tool that needs to be avoided as it can upset the duality function of the joint-stereo parameter in the encoder. The stereo sound field should be left as it is. Any stereo image manipulation for effect should be tested in the main audio-chain processor, if at all.

There are many sophisticated dynamics boxes and plug-ins available today, and they work extremely well, unto themselves. But upon being added to the HD Radio signal path, they can create havoc. Moderation is the key.

Just as practice and experience aided in determining how much to process in the production room for conventional broadcasting, the same holds true here. The adage "less is more" rings true now more than ever.

Content Quality: This is significant as it plays into another old adage: "garbage in is more garbage out."

It's imperative to ensure that content is of pristine quality. Music and production libraries need to be of the highest possible quality. Any production or music service that provides content that has been independently sweetened should be avoided, as it's not really known what that sweetening was. Audio sources should be linear as much as possible. This reduces the amount of transcoding that occurs when multiple bit-reduced signals are passed.

When transferring audio sources, ripping CDs or dubbing elements, make sure that the audio I/O ports are capable of providing professional-grade audio quality. Many sound cards or computer audio interfaces are not designed for high-quality professional audio; sonic degradation results. This transforms into distortion and added artifacts in the HD Radio signal (and lesser-grade audio in the conventional signal).

Transcoding: This is the process of passing one bit-reduced audio signal onto another codec. It can be done with minimal degradation when the initial source is bit-reduced at a higher rate than the final coded pass. Because the FM system uses 96 kbps, any prior coding should be at least 256 kbps or higher.

This is a topic that is unavoidable in some instances - for example, the remote feed that arrives via an ISDN or POTS codec. In those cases, the source audio should be totally unprocessed, as it will be passed onto further data reduction at the transmission point. Any EQ should be kept to a minimum. These instances exist and we must deal with them. When source elements can be controlled, eliminate all bit -- reduced content from hard -- disk servers or play-out systems. Using a high-quality linear source will generate better performance, as transcoding is eliminated.

Using MP3 files should be used only in the event that a linear source is not possible at all. If an MP3 source is required, the bit rate should be as high as possible. The suggestion here would be anything above 256 kbps.

CDs ripped down to MP3 files should be avoided at all costs for music programming. MP3 files for news content are not as much of a problem as they usually involve only a voice element. But commercials, promos and imaging elements should be produced in the linear domain, without exception.

Naturally, as HD Radio continues to roll out and evolve, there will be plenty to gather from real-world experience. Products and applications are under development for producing in the data-reduced world. What was sonically possible 10 years ago in Codec Land has been surpassed by leaps and bounds. The progress doesn't seem to be slowing down, either, thereby allowing further improvements to the system as a whole.


Image placeholder title

HD-R Delay Needn't Be a Pain

In order to make possible a seamless blend, broadcasters are now forced into delaying their analog signal in order to match the digital signal. When broadcasters implement this delay, there are several considerations that come into play.