Public Okay With Lower AM Bandwidth?
Over the past two years, the AM Broadcasting Subcommittee of the National Radio Systems Committee has been studying the effect of reducing bandwidth on AM transmission systems, trying to determine answers to several questions:
| Figure 1|
| Figure 2|
| Figure 3|
- What is the current state of AM receivers in the general marketplace?
- Do today’s analog transmission standards and limitations make the best use of the AM allocations?
- What does the listener perceive if AM transmission bandwidth is changed?
- Can we make changes that will reduce AM tune-out?
The AM Study Task Group, led by Omnia President Frank Foti and NPR Labs Senior Technologist John Kean, studied all of these questions and more. The results of that study are summarized here and the full report is available on the NRSC Web site at www.nrscstandards.org.
The AM Broadcasting Subcommittee is now turning its attention toward the review and possible modification of the AM transmission and reception standards known as NRSC-1, -2 and -3. We request that interested persons join us in that task.
NRSC-1, -2 and -3 cover, respectively, AM broadcast preemphasis/deemphasis and audio transmission bandwidth; emission limitation for AM transmission; and audio bandwidth and distortion recommendations for AM receivers.
Meetings are held via telephone and are open to any interested parties except the press. The AM Broadcasting subcommittee requests and encourages your participation in this important matter.
By Jeff Littlejohn
The author is co-chair of the NRSC AMB Subcommittee and executive vice president of distribution development for Clear Channel Radio.
The AM Broadcasting Subcommittee of the NRSC was formed in 2004 to maintain NRSC standards relating to analog AM broadcasting. The subcommittee is reviewing three standards: NRSC-1, NRSC-2 and NRSC-3. The subcommittee can reaffirm, modify or retire these standards.
Members noted that some broadcasters have already reduced the bandwidth of their analog AM signals from the 10 kHz specified by the NRSC standards to 5–6 kHz, in an effort to reduce the interference in the band, and with the understanding that most consumer receivers are band-limited to 5 kHz or less. The group considered a proposal that that the NRSC consider reducing the bandwidth specification in NRSC-1, -2, and -3 to something less than 10 kHz, but the subcommittee agreed that before such an action could be considered, it should study analog AM receivers, characterizing, among other things, receiver bandwidth, and consumer reaction to reduced bandwidth.
In late 2004 the Subcommittee formed the AM Study Task Group to determine whether consumers would reliably perceive the audio quality differences of AM transmissions at various bandwidths, recorded through commercially available receivers, and whether these perceptions would affect consumers’ continued listening behavior. The AMSTG subsequently conducted a consumer subjective evaluation study of audio obtained from three prototypical receivers, as well as an objective evaluation of audio performance of a large number of current consumer analog AM receivers, including OEM and after-market car radios, shelf mini-systems, boom boxes, table radios and portables.
Before considering potential changes to the NRSC-1, -2, and -3 standards, the AMB is sharing the results of the AMSTG study and is seeking input from all interested parties. The following summarizes the methodology used and findings obtained from the AMSTG study.
AM receiver measurements
NPR Labs completed objective measurements of 30 consumer analog AM receivers in late 2005 with support from NRSC co-sponsors CEA and NAB. These laboratory measurements collected data in two areas:
- Baseline audio performance of the receivers, including frequency response, harmonic distortion, intermodulation distortion and signal to noise ratio;
- Objective noise level differences with signal interference at several audio transmission bandwidths (i.e., 5, 6, and 7 kHz), relative to the current transmission bandwidth standard of 10 kHz. Weighted quasi-peak noise measurements were taken to approximate the response of human hearing to audible noise. A first-adjacent channel (±10 kHz) interfering signal was modulated with a pulsed frequency-shaped noise to simulate the characteristics of program audio.
These objective measurements established that the majority of current analog AM receivers have audio bandwidths of less than 5 kHz. In fact, with only a few exceptions, the frequency response of individual receivers falls off above 1 or 2 kHz.
As shown in Fig. 1, the combined frequency response of all receivers through the test bed (the middle curve, in blue) was –3 dB at 2450 Hz and –10 dB at 4100 Hz.
The overall variation in audio bandwidths was wide, as shown by the standard deviation for the entire test population (+1s in green and –1s in brown): at 4100 Hz, the first-order standard deviation was approximately –2.6 dB and –17.2 dB, a range of 14.6 dB. The table inset in Fig. 1 lists the –3 dB and –10 dB bandwidths for the receivers by category.
Further, each receiver was evaluated for change in noise at the audio output with first-adjacent channel interference using audio transmission bandwidths of 5, 6, 7 and 10 kHz at desired-to-undesired RF signal ratios of 30, 15, 6 and 0 dB. The effect of transmission bandwidth on weighted quasi-peak SNR for the combined receivers is summarized in the Fig. 2, showing that reduced transmission bandwidth offers SNR improvements of up to 12 dB, relative to 10 kHz bandwidth, with first-adjacent channel interference.
Industry, consumer testing
Based on the findings of the receiver measurements, Sheffield Audio Consulting and NPR Labs conducted subjective testing between February and May 2006, using audio recorded from three of the tested receivers. As a practical matter the number of bandwidths to be tested has to be limited. The AMSTG decided to use three bandwidths: 10 kHz, the current NRSC standard bandwidth and maximum bandwidth allowed under current FCC rules; 5 kHz, which represents the maximum bandwidth where adjacent channels do not overlap; and an intermediate bandwidth.
To establish this intermediate bandwidth, NPR Labs conducted a “phase 1” listening test that included 18 broadcast industry representatives and determined that 7 kHz was the best intermediate bandwidth to be included in the consumer test.
In the “phase 2” listening test, consumers judged (a) which transmission bandwidth, 5 kHz, 7 kHz or 10 kHz, had the best quality, (b) the magnitude of the difference between the quality experienced using these bandwidths, and (c) whether they would continue to listen to the audio, given the quality of each of the samples. Audio samples used in this test included those impaired by first-adjacent channel interference in addition to unimpaired reception.
Audio source material consisted of material supplied by NRSC music test samples and NPR speech samples and Greater Media sportscast and commercials.
Forty-four listeners, distributed between ages 19 and 71, participated in the consumer test. Data from 40 qualified listeners — 20 female and 20 male — was collected.
From the pool of 30 receivers that were objectively tested, NPR Labs chose three from which to record audio samples: the JVC KS-FX490 car in-dash cassette (median-bandwidth); the Panasonic CQ-CB9900U in-dash CD/HD Radio (80th percentile bandwidth) and the Aiwa JAX-S77 portable boom box (20th percentile bandwidth).
An A/B comparison was used to obtain listener judgments. Test participants listened to seven different samples, including female and male speech, voice-over (commercial), a sportscast, and rock, country and classical music.
Over the course of the entire test, participants listened to a total of 189 sample pairs. After listening to each sample pair, consumers judged which sample they thought had better quality, how big the quality difference between samples was and whether they would continue to listen to the audio for either or both of the samples.
The graphs in Fig. 3 show the percentage of participants who picked one bandwidth over another, such as 5 kHz over 10 kHz, at various D/U signal conditions, separated by genre and aggregated for the three receiver bandwidths tested: 20th percentile, median and 80th percentile bandwidth.
Notice in Fig. 3 that the findings for “speech” follow a significantly different pattern than findings for all other genres. Participants clearly favored 5 kHz and 7 kHz in speech, while in music and commercials preferences were not as clearly articulated. For sportscasts, participants demonstrated a slight preference for higher bandwidths under less impaired conditions.
Consumer subjective test results suggest the following:
- For music, commercials and sportscasts, little difference was heard between 7 kHz and 10 kHz bandwidths, regardless of first-adjacent channel interference conditions. For speech, which does not mask noise and interference, larger differences were perceived, based on impairment conditions;
- In unimpaired or moderately impaired conditions as determined by the desired-to-undesired signal ratio, people tended to prefer higher bandwidths to lower bandwidths. However, 7 kHz and 10 kHz bandwidths had equal preference;
- With speech in moderate to heavy impairment conditions, participants preferred lower bandwidths, 5 kHz and 7 kHz, to higher bandwidths, despite a mutual reduction in transmission bandwidth on the desired channel.
In the majority of listening conditions, consumers preferred either 5 kHz or 7 kHz, and often reported that 7 kHz was equivalent to 10 kHz in unimpaired or moderately impaired conditions. These preferences were articulated most strongly in speech conditions, where noise from interference affected listeners the most.
In extrapolating this consumer data to general public listening, it is important to note that discerning background noise is easiest in speech conditions, and thus the speech testing represent the most critical results. This is important for two reasons: (a) the majority of AM programming includes speech, and (b) consumers will hear more noise in any music, sports, and commercials that are qualitatively less “dense” than the programmatic material included in this test.
Since consumers seem to be most critical of “noise” and seem to tolerate more constrained bandwidth when they receive a clean signal, it’s likely that lower bandwidths will satisfy consumers in most conditions.
The AMB Subcommittee welcomes additional participation and comments from interested parties. Read the full report at: www.nrscstandards.org/AMB/default.asp .
RW welcomes other points of view. Write to firstname.lastname@example.org.