A DOE Relates His Experiences Upon Listening to ‘Real-World’ HD Radio for the First Time.
A couple of weeks ago, I installed a Kenwood KTC-HR100MC HD Radio tuner in my truck. The installation was an easy one-hour project. In fact, the whole experience, from order to radio-on, was pleasant and frustration free.
I ordered the unit from Crutchfield. The sales rep was knowledgeable, up on HD Radio and the different units that were available. He also provided, free of charge, the wiring harness adaptor and mounting plate for my specific make and model of vehicle along with a good set of instructions that clearly showed how to get into the dash to remove the old radio, how to mount the new unit and how to wire it into the existing harness.
There was zero guesswork, and the finished installation was professional and complete. The new radio looks like it belongs there.
Fun
Since I installed the new radio, I have had a lot of fun dialing around the AM and FM bands, listening to the HD Radio stations in the Denver market.
Denver is not all that well populated with HD Radio signals, but there are enough to get a feel for what this new digital terrestrial medium is all about. There are four FM stations and five AMs transmitting HD Radio signals that I can receive locally.
(click thumbnail)The author tunes in on his new Kenwood HD Radio receiver.
Listening to “real-world” HD Radio for the first time, I found that on FM, there was no “night-and-day” difference from analog FM. When the analog blends to digital a few seconds after tuning a station, there is a perceptible difference, notably a complete absence of noise and a much more “open” sound, presumably the result of lighter processing and the absence of clipping. All the local HD Radio FM stations have fairly well matched the audio levels of their digital and analog signals, and all but one are in perfect time alignment. The one station that wasn’t perfectly time aligned was out by maybe 50 mS or so, enough to cause an “echo” effect during the blend-to-digital transition.
AM, on the other hand, really is night and day. As the blend-to-digital takes place, the high and low ends open up, the noise disappears, stereo separation becomes apparent (in some cases; some of the talk stations were transmitting in mono) and the audio sounds like noise-free FM.
No doubt about it, AM stands to gain the most in quality when converting to HD Radio, although some might argue that FM will be the real winner with multiple audio streams.
Artifacts
One negative thing I have noticed on some of the local AM stations, however, is the presence of digital artifacts. It’s apparent on some of the stations all the time, and most of the stations some of the time.
One station sounds truly great, a local independent Christian station on 910 kHz. There is no artifacting, and especially during music segments, the station sounds outstanding.
I hear digital artifacting just about all the time on Denver’s big 50 kW talker, and I hear it during at least some segments on all the other AMs except the one on 910. This has me wondering what the cause is. A few things come to mind: source material (MP2 files on a digital automation system?), profanity delay, digital STL and processing. I suspect that the 910 station sounds good all the time because it plays CDs, uses no STL (studio and transmitter are collocated) and processes lightly.
Source material certainly has the most potential to cause trouble, sometimes having been through the MPEG grinder one too many times, or worse yet, through a non-MPEG compression cycle and then through the MPEG grinder.
I have no idea what degradation is caused by a profanity delay when combined with multiple compression passes, but I suspect it’s something we’ll have to figure out. Digital STLs would seem to be an area of potential trouble, particularly the Moseley DSP6000 digital transmission system. That system uses ADPCM coding at a 32 kHz sample rate. You can avoid the ADPCM coding if you go in AES, but you’re still limited to 32 kHz, which presents a significant bottleneck.
In my experience, sample rate down-conversions are okay as long as you stay at the reduced rate. The trouble is, most HD Radio transmission equipment operates at 44.1 kHz, so an upconversion is necessary, one that necessarily involves interpolation of data. Presumably with the Starlink, one would downconvert to 44.1 kHz at the studio and stay at that rate all the way into the HD Radio generator.
Digital processing also presents an opportunity for degradation and artifact production, particularly if a sample rate up-conversion takes place in the processor. It is my hope that these stations’ engineers will listen to their stations on a digital tuner and work to clean up the artifacts.
Although our stations do compete with these stations, we all lose if the general listener impression is that AM HD Radio sounds like a fair Internet stream. We all win if we all sound good.
You can be sure that our Ed Dulaney will be paying close attention to sound quality on KLZ, KLTT and KLDC when those AM stations begin HD Radio transmissions later this year. I suspect that significant changes in the audio chains are forthcoming.
Lessons
One thing is for sure: We’re all going to have to find ways to avoid multiple MPEG passes and sample rate conversions in the future. This means standardizing on codec settings, choosing the most advantageous sample rate (no doubt 44.1) and sticking with it. It also means getting into the digital domain and staying there.
That may be easier said than done.
Consider the typical case where the satellite antenna is located at the transmitter site. Let’s count the A/D conversion cycles.
Satellite feeds are demodulated and output as analog audio (that’s one). The analog satellite audio is fed to the Intraplex PT- 255 and fed to the studio where it is decoded to analog audio (that’s two). The analog audio is mixed in the analog domain and fed through an Eventide BD500 profanity delay (that’s three).
The analog audio out of the delay is then fed to an AGC and then back into the Intraplex PT-255 and on to the transmitter, where it is decoded back to analog audio (that’s four). That analog audio feed is then fed to the input of the Omnia processor. In analog-only stations, the analog output of the processor is fed to the transmitter audio input (that’s five). In digital stations, digital outputs are fed to the HD Radio generator, and an analog output is fed to the audio input of the transmitter (again, that’s five).
So in a typical case, we’re looking at five A/D cycles, some of which may have been at different sample rates and at least three of which involved MPEG coding. Should we be surprised that there are digital artifacts in the on-air signal?
If our studios were all-digital, we could cut a number of cycles and all but one MPEG pass.
The satellite audio could be taken as an AES digital feed (hopefully at 44.1 kHz) and fed to the studio as AES via a linear AES card in the Intraplex. At the studio, the AES satellite feed is connected directly to the bridge router or digital console, mixed and output as 44.1 kHz AES. That signal is run through the profanity delay as AES and then fed back to a linear Intraplex AES module for delivery to the transmitter site.
At the transmitter site, the 44.1 kHz AES would be fed directly to the Omnia and processed. The AES outputs would be fed directly to the HD Radio generator, and an analog output of the HD Radio gen (there’s one) would be fed to the audio input of the transmitter.
All that boiled down to just one A/D cycle, and no trips through the MPEG grinder. So what’s stopping us from making those changes now? Not much – just cost. Our entire infrastructure in most of our AM stations is set up for analog transmission. It’s going to take time to swap out analog boards and modules.
But even if we can’t immediately make the entire path digital, we must find ways to eliminate A/D cycles and MPEG passes. For every one we eliminate, we’ll make a measurable improvement in the station sound. Be thinking about that as your date with HD Radio approaches.
RW welcomes accounts of your own HD Radio listening experiences.
