What special considerations in audio processing come into play for radio stations that deploy HD Radio?
We asked three experts for this article, which is excerpted from the free ebook “HD Radio Best Practices 2025.”
“First of all, 90% of your HD listeners are going to be in vehicles,” said Mike Pappas, vice president of global sales for Orban Labs.
“This means that as you set up your processing, you will need to be inside a vehicle to ensure the proper sound and processor performance. And as every vehicle’s sound system is different, you are going to have to listen in multiple vehicles.”
He said there’s a finite amount of HD Radio bandwidth to be “divvied up” between all of the HD channels you plan to run.
“Otherwise, you’re going to run into limitations on the number of bits for each of your HD channels. The latest release of the HD Radio codec Extended Hybrid Mode MP11 gives you 148 kbps to use, but requires significantly more transmitter headroom.”
Frank Foti, executive chairman of the Telos Alliance and founder of Omnia Audio, said the most important aspect is to ensure that the audio processor is designed for use with data-reduced audio, which is the transom that HD Radio employs.
“Digital radio is basically streaming audio that is carried over the air. Most commercially available processors provide dedicated processing for HD Radio and streaming services, as well as FM stereo,” Foti said.
Unlike over-the-air FM, audio processing for HD doesn’t need to include algorithms for FM pre-emphasis or the radical peak control techniques, according to Jeffrey Keith, senior audio processing development engineer for Wheatstone.
“That means that properly done, a station’s HD signal can sound just like a bigger version of the source CD as far as audio quality is concerned.”
Keith said that most on-air audio processor manufacturers today provide competent HD processing among their offerings.
The main signal
To make FM+HD1 stations sound as good as possible, Pappas said, assign a minimum of 48 kbps to the HD1, then work the “blend” between the analog FM and the HD1 to make the transition more natural.
“That means the apparent levels should be as close to each other as possible. Sometimes reducing the HD1 HF bandwidth is needed along with EQ and processing changes. Again, listen in lots of different vehicles in order to make the best choices. You don’t want to knock listeners out of their cars when their radios transition to HD1 from analog FM and vice-versa.”
Modern processors, said Frank Foti, provide dual paths that support both conventional FM analog, as well as HD Radio transmission.
“It is imperative that the two paths are ‘sample matched’ so the blend from analog to digital is a seamless transition,” Foti said.
“Additionally, processing for conventional FM and HD requires a bit different approach. Conventional FM utilizes a preemphasis boost to the high frequencies and hard clipping for absolute peak control, whereas the audio signal for HD Radio is a flat response and peak control is accomplished using a look-ahead limiter.”
He said both peak limiting methods behave differently, and sonic side effects from each sound different as well.
“On account of that, one must set up processing for those respective methods. A good understanding of these methods and how they sound will assist greatly in achieving the final result.”
Multicasts
What about processing for HD2, -3 and -4 channels?
“The HD+ channels, as I like to call them, should utilize the same style of competent audio level and peak control as is offered in the HD section of on-air processors,” said Jeff Keith.
Foti encourages engineers to employ a processor dedicated for streaming audio and that will support low bitrates.
“There are software-based processors that are both applicable to and cost-effective for this application. Many of them provide functions designed to improve audio quality at lower bitrates and minimize coding artifacts.”
Pappas recommends that users apply a deft touch with compression and limiting; he believes “less is more” when it comes to low-bandwidth HD channels.
“Additionally, at very low bitrates you might find that dropping the HF bandwidth down to 12 kHz or 13 kHz reduces the amount of HF artifacts. And if your HD2/3/4 signals are running ‘talk,’ mono might be an option to check out, or parametric stereo. (Note that in Gen3, stereo is the default codec for these channels, but in Gen4 parametric stereo is the default. If you have older Gen3 Capture Clients running on a Windows-based Importer, it’s best to make sure they are all running parametric stereo or mono as appropriate.)”
The point is you should not be afraid to experiment.
“Good-sounding mono is better than horrible-sounding stereo. At the end of the day you don’t want this to sound like ‘satellite radio,’ which has a less-than-stellar reputation for quality,” Pappas said, reiterating that you should listen in multiple vehicles to make the best processing choices.
New features
These experts say the technology has improved quite a lot.
“Processing has been evolving, and what’s available now is significantly better than what was available even five or six years ago,” Pappas said. “If your current processing is old enough to get a driver’s license, you owe it to yourself to audition a current processor.”
Keith said Wheatstone’s offerings include a separate final processing stage for the HD side of things. “We split the audio into separate FM and HD final processing paths after the front-end multiband AGC and limiting, and we make that split there so that blend back and forth between the HD and analog signal can be completely unnoticeable to a listener.”
Foti said the performance of data-reduced audio is affected by both intermodulation, or IMD, and harmonic distortion, or THD.
“Employing a processor that is not only designed for HD transmission but also contains processing algorithms that yield low levels of both aforementioned distortions is imperative, the reason being that distortion products from processing can aggravate coding artifacts, even at higher bitrates,” he said.
“For example, we have identified that processing operations that make use of fast attack/release settings will exaggerate coding artifacts, as faster time constants tend to add IMD to the signal.”
Comment on this or any article. Email [email protected] with “Letter to the Editor” in the subject field.
