Despite this, Yonge said that after an initial flurry of interest things "went quiet for a bit". Even so the following years did see some movement; Digigram introduced an Ethernet protocol in 2001, following the standardisation of Gigabit Ethernet at 1000Mb/s in 1999, and the AES itself launched audio over ATM in 2002, building up to the introduction of Dante in 2006. This audio over IP (AoIP) system is also based on Ethernet and was developed by Australian company Audinate primarily for the live sound sector, although it has begun to move into broadcast.

Two years later Philipp Lawo, chief executive of the console manufacturer that bears his last name, began discussing the possibility for an AoIP system based on open standards rather than the more proprietary and licensed approach of Dante. The result of internal R&D and collaboration with industry partners including AEQ, Merging Technologies and Genelec was Ravenna, launched squarely into the broadcast market at IBC 2010. Andreas Hildebrand, senior product manager at ALC NetworX, which develops and markets the technology, describes Ravenna as a future-proofed replacement to ATM (Asynchronous Transfer Mode).

Also in 2010 the IEEE (Institute of Electrical and Electronics Engineers) published 802.1 for interconnection, networking and data centre bridging, carrying, routing and switching audio and video signals over standardised Ethernet connections. This forms the basis of AVB (audio video bridging), a non-manufacturer aligned, open standard based on compliant switches - the bridge - to connect different types of equipment. The technology is being developed and adapted for three defined markets: professional AV (including broadcasting), automotive and consumer. Rick Kreifeldt, president of the AVnu Alliance, the industry forum behind AVB, comments that the intention is to "try to reduce complexity in the network".

The Contenders

AVB, Ravenna and Dante are the three leading contenders in the race to produce a platform for networks carrying high quality, multi-channel audio. Speaking at the Audio Networking Forum, Mark Yonge said that the aim now was to achieve networking not for file exchange, which he said was "pretty much known", or voice over IP, but for streaming audio over a network. This, he emphasized, meant "production grade audio, PCM at 48kHz and 24-bit, or better, for broadcast, music or live sound".

Lee Ellison, chief executive of Audinate, says that Dante meets such standards because it was originally designed to meet "the exacting demands of live sound". It is also scalable, he adds, so is now being used in broadcasting. The move to networking is being driven in part by the increasing use of IT-based technologies and working practices in TV production; while this is bringing greater efficiency and lower installation costs, it also highlights the existing difference between computer operations and traditional broadcasting engineering.

Mark Yonge notes that while broadcast engineers and operators are now familiar with computer-based systems, there is a still a need for IT technicians to be "media capable" and handle all forms of broadcast material, including audio.

Dante technology is licensed to manufacturers and developers who integrate its networking capability into their products. Among those already signed up are Yamaha, Stagetec, Studer/Soundcraft, NTP, Cadac, Zaxcom and broadcast console company Calrec Audio. Commenting on the licensing and proprietorial aspects of Dante, Ellison says, "Developing a standard can be enormously cost prohibitive for a company, especially producing the proper interfaces, but that's what we deliver. The idea behind Dante is that it discovers all the devices on a network and makes setting up something like an intercom easy."

Ravenna

By contrast Ravenna was conceived as an open system that did not leave the design of the user interface in the hands of any one company or developer. "This allows manufacturers to set up the system in application specific ways," says Andreas Hildebrand. "That makes us different form the major player in the AV market [Dante]. Basically all those devices are controlled in the same way. We've left that totally open because we didn't want to force people into one way of looking into the system." While this remains the aim behind Ravenna, ALC NetworX is working on a simple user control unit.

Where Ravenna is intended as a replacement for the now outmoded ATM protocol, AVB is seen by its developers as a successor to SDI, which still forms the backbone of many broadcast facility installation today. Rick Kreifelat comments that the challenge for this technology in audio and video is creating big networks handling a lot of activity. "Automotive networks are easier to understand than those in AV are because they are smaller," he explains. "In pro audio and video you're dealing with large systems where devices are coming on and off the network."

Among pro AV members of the AVnu Alliance are Harman, Avid, Barco, Shure and Sennheiser but the common thread is the switching technology, which is being developed by a handful of specialist companies. The key to an AVB network operating correctly is having all equipment connected to it complying with the standard and interfacing with the switch. This calls for rigorous - and expensive - testing of products, carried out by approved test houses. This, and the design, is complicated and, says Kriefelat, costs "hundreds of thousands of dollars". It is, he adds, something the AV industry has not seen before but the aim is for buyers and users of equipment to see the AVB certificate and "know that devices will work together", something that will bring more options in networking. "But you personally do not need to understand the certification process," he says.

In a presentation on AVB to the Audio Networking Forum, Jan Eveleens, chief executive of Axon Digital, outlined that the extensions added to Ethernet by the IEEE technology offered mechanisms that overcame the "best effort delivery" nature of the carrier, offering more of a guarantee of when data will arrive; was not random in but more deterministic; introduced a sense of time and/or timing; and was aware of different types of material.

He observed that audio was 75 percent of the maximum data rate in an AVB network, with the remainder being made up by other data. He later commented that there is "substantial potential in broadcasting as the existing SDI-based infrastructure has come to the end of its life cycle and customers are starting to look at (networked) alternatives", particularly as 10Gb/s (and higher speeds) were now very common and affordable. "We not necessarily need IP (layer-3) functionality as most of the live production applications in broadcast are pretty local, such as inside an OB-van or studio complex," he said, "so layer-2 is OK. Would it be great if AVB layer-3 were available? Of course - and this is being worked on - but it's not essential."

Time-Sensitive Networking

Towards the end of 2014 the AVnu Alliance added an industrial strand to its remit and announced development of a second-generation version of AVB, known as TSN (Time-Sensitive Networking). This caused speculation that AV - and audio in particular - would be marginalised. Eveleens responded, "The AVB/TSN community is very broad and very active. So, while work is on-going on TSN, there is equal sized efforts in AV oriented applications. The work in the revision of that standard is almost entirely focused on AV applications as a large set of new AV transport formats were added. If a standard does not get traction in a certain industry it could become marginal but it is much too early for that. Real-time audio networking has been on-going for several years, as it is about 1000 times 'slower' than video and thus requires less bandwidth and less advanced technology. In video it is just starting now, so it is really hard to predict what technology will get into pole position and will win the race. We are just in the early phase where all players do some early trial runs."

Technologies are now maturing and products are appearing on the market but the actual amount of streaming networking going in broadcasting appears to still be low. Audio consultant Roland Hemming told the Audio Networking Forum that there was no definite way to determine how much sound is being networked right now.

The situation, he added, was not helped by "skewed data". A recent survey showed that while 58 AVB-compliant products were on the market, 32 were from just two manufacturers. This applies to other formats, with 25 of 104 CobraNet amplifiers being made by one company.

The possible outcome of the present situation, Hemming concluded, was that "protocol fiefdoms" could be created, with each format being selected for its benefits in a specific application. This could turn out to be Ravenna largely for broadcast, while in live sound it might be Dante handling front of house and monitors and Q-Lan on back of house. A significant question, Hemming pondered, was if AES76 will the connecting factor between the different formats as "the lowest common denominator".

Published in September 2013 this was developed by the AES, with input from the European Broadcasting Union. It is intended as an engineering standard for networked/streaming audio over IP, described as "an interoperable subset of existing technologies" designed to further the development of AoIP, rather than be a new technology in its own right. This has been overlooked by some in the industry, who have taken to lumping AES67 in with Ravenna, Dante and AVB as another networking standard rather than a way to form networks by interfacing between different formats.

There is further potential confusion from the use of manufacturer-specific routing and distribution systems as part of an overall network. These include Riedel's MediorNet, a real-time networking system for signal transport, routing and processing and conversion of video, audio, data and communications; the RTS OMNEO media networking interface card system, which converts the ADAM intercom system into an IP-based, AVB-compatible communications network using AVB for media control while control functions are carried out in OCA (Open Control Architecture), with everything carried over IP transport; and Calrec Audio's Hydra 2.

This is the latest incarnation of Calrec's routing and networking system for its digital consoles. During the Audio Networking Forum Patrick Warrington, technical director at Calrec, commented that people were now looking at the networking of not just audio but video and communications as well on a single synchronised platform. The challenge was, he continued, that there are several "layers" to a broadcast event, starting with the central production or acquisition, going through the supporting infrastructure of editing and post to distribution. This could involve numerous formats, including SDI and MADI (Multichannel Audio Digital Interface), which is now being used widely for audio in broadcasting, and not necessarily rely solely on AoIP.

The possible future, he said, could be "hybrid networks", featuring something like a Hydra 2 at the centre running MADI, with an interface to an IP system for distribution. Such an approach has already been taken by UK broadcasters BSkyB and BT Sport.

The next phase of networking will involve more control elements, something the AES is looking into for AES67, all of which implies that audio networking will continue to be a complex and evolving area for some time to come.