Skip to main content

Care And Feeding Of Audio Networks

The Audio Engineering Society Technical Committee on Network Audio Systems (TC-NAS) was chartered in 1998 when it was recognized that the major increase in network bandwidth and quality of service made it possible to consider brand new classes of audiovisual applications, as well as unconventional and high-performance implementations of existing ones.

In planning a proposed tutorial covering the care and feeding of audio networks, the committee recognized that there are now numerous network and audio technologies and numerous real-world applications for these technologies. While the combined expertise of a panel of professionals would surely cover much of the material, the committee recognized that backing up expert testimony with an online survey would increase coverage.

The overall intent of the survey was to identify current applications for audio networks and understand requirements imposed by those applications.

The interesting result from the survey was a strong message of the desire to see open and multi-sourced technology standards for audio networking, as no standard currently exists. The survey reminded us of the core importance of control and monitoring capability on a network. The survey uncovered divisive opinions as to whether an audio network is best built as a standalone infrastructure or better to make use of data communications infrastructure already planned or in place in many facilities.

One of the original hypotheses I held for this survey appears to have been disproved. I had assumed that different applications would call starkly different requirements; the survey showed only minor differentiation.

We had three categories: Overall (114 responses), Installed (26) and Live (30), each with sufficient respondents to draw responsible application-specific conclusions.

AES Workshop
The survey results were presented at the 2005 AES Convention. As the chairman, I chose from among the many highly qualified survey respondents and invited a panel of system designers and users to help interpret and discuss the results. Dave Revel and Roland Hemming represented installed sound applications for stadiums, theme parks and the like. Bruce Olson covered theater sound. Jim Meyer and Jeff Berryman are both system designers for concert sound companies, Clair Brothers and Jason Audio, respectively. James Yeary from Turner represented broadcasting interests. Consumer applications drew only five responses. Nevertheless, I invited Tom Blank, a researcher at Microsoft, onto the panel.

In our global market-centric business climate, engineering initiatives now tend to attract R&D funding proportional to their revenue potential. High-volume consumer applications are receiving proportionately greater attention, and the result is that technology development often flows from consumer implementations to the professional applications.

Respondents were asked how they evaluate the cost of an audio network. Installed-sound people report they are twice as likely to make their cost comparisons against other audio network technologies, whereas in live sound the situation is reversed, with respondents twice as likely to compare cost with analog systems.
Overall, respondents were evenly divided between analog and network cost comparisons. Only a small number of respondents make reference to the third survey option: existing digital distribution technology (i.e. AES3 with central routing).

A possible reading of the reported marginal role of existing digital technologies is that those considering using an audio network are currently using analog technology.

The survey asked two questions regarding audio networking standards. Respondents, regardless of application area, overwhelmingly supported "open," "multi-sourced" technology. The reported demand for open, multi-sourced standards is an interesting situation given that, at this point, all the technologies used in respondents' applications areas are proprietary or, practically speaking, single-sourced.

The panel suggested that the desire for open standards was probably due to the fact that "open" is a more palatable word than "proprietary." The real requirement is for a widely available interoperable network. Those requirements can be met by either open or proprietary technology with fair licensing terms.

A generous portion of the survey was concerned with the issue of latency. Overall, half the respondents, in negotiating latency issues, report that they look to a comparable analog solution as a reference. The remaining respondents were split between working in non-latency-critical applications and those who take the particulars of the situation into consideration.

Among the installed-sound responses, it was not surprising to find that the analog benchmark was a less frequent choice. Instead, more installed respondents reported latency as not a critical factor, and an application-specific latency criterion was the most popular response.

In all application areas, the most common absolute latency requirement cited was "less than 5ms." "Less than 3ms" and "less than 1ms" were also popular responses.

Overall, stage and in-ear monitoring (IEM) were the most frequently cited latency-critical scenarios. Synchronization to video was another frequently mentioned scenario.

Panelists Berryman, Revel and Meyer were in general agreement that 5-15ms was the happy zone for the more critical IEM scenarios.

Distance And Extent
The applications respondents spoke to were, by and large, local area network to campus-sized installations. The most common distance requirement for live applications was 5-100 meters. For installed sound, the most popular choice was 100-2,000 meters. There were very few takers for cables 5 meters or shorter. Professional audio is apparently not giving much consideration to audio networks for in-rack wiring.

There was good acceptance of twisted-pair, coaxial and fiber-optic cables. Reflecting the low turnout in consumer applications, plastic optical fiber (POF) registered as the least appreciated cable type. Perhaps surprisingly, Cat-5 beat out analog shielded twisted-pair (STP) as the comfort cable for audio professionals.

Installed sound indicated the highest degree of comfort with optical cable though half the respondents in the other categories were also comfortable with this medium.

The survey showed a tension between dedicated and shared network design practice. Overall, 49 percent indicated a preference for a separate audio infrastructure. 42 percent believe there is value in convergence, and only 12 percent showed flexibility, citing the issue as not critical or critical only in terms of cost.

When broken out by application, a greater percentage of the installed-audio respondents were in the convergence camp, reflecting a push towards integration of systems in new installations.

The panel's read was that this division in audio network design approach is quite real. Some designers, in the interest of quality and/or reliability, work to remove dependencies on (data communications) systems outside their control. Others buy into the promise of convergence and use it as a selling point for their forward-looking designs. Even if the audio designer doesn't have a lot of skin in this design decision, it's quite possible that his client or his client's client does.

Another divisive dimension is that today's audio networking technologies and the marketing behind them is divided in approach. EtherSound, A-Net and AES50 to name a few, require a dedicated infrastructure and, in most cases, tout that as an advantage. CobraNet, AES47, mLAN and Axia, to name a few more, are designed to work on a shared network and tout that as an advantage.

Monitoring And Control
Networks have a legacy of association with control and monitoring in audio systems. Given this, it should not be a surprise that almost all respondents expressed a desire for some level of control and monitoring capability (only four responses indicated that none was required).

Control and monitoring is an expected capability, and this may be an important factor driving adoption of audio networks. It appears that, especially in installed sound, the primary draw of an audio network may be in the control and monitoring capabilities. The ability to distribute audio over the same network merely makes it a more attractive solution.

77 percent of responses specified a requirement for no audio interruption or automatic recovery from failures. The call for reliability was most striking in live sound where more than half of respondents requested no interruption of audio in the face of hardware failures.

The panel cited emotion as the driver for these requirements. Audio networking, and digital audio, is less accessible to system operators. A sound engineer is likely to be more comfortable and successful correcting a ground loop than a network router misconfiguration.

There is no doubt that digital audio technology will continue to edge out analog, and with that, there is little doubt that networking will be the preferred distribution method. Our overall goal should be to ensure that audio networking evolves into an interoperable and cost-effective technology that solves problems of our applications.