The back of the net

Published: August 02, 2010 ASIA

Audio networking is transforming the sound reinforcement scene, but what’s the ultimate goal? Phil Ward investigates

A computer network is simply a collection of digital devices that talk to each other, so that the users or devices in it can share resources and information. When audio systems were analogue, they did exactly that – every signal path wound its merry way from transducer to amplifier and loudspeaker channel by channel, and the output sound was its own data. Nothing crashed, nobody swore and it all happened at the speed of sound.

Then the Earth cooled, dinosaurs became extinct and pro audio went digital. Suddenly the audio system is a computer system, and the amount of data on tap has expanded to occupy the bandwidth available for it. Few would claim that the handicaps outweigh the benefits, but there’s no doubt that networking has become an industry marketing buzzword. And buzzwords, as Stephen Hawking almost said, are the enemy of knowledge.

Working titles

Most of the current digital audio networks are based on Ethernet, the Cat5 cable and RJ45 connector-based IT standard for local area networks. Its popularity is no doubt based on its ubiquity, but there are those who question its ultimate suitability for high-end pro audio – just as CD has never really satisfied audiophiles as a replacement for vinyl.

First out of the block was CobraNet, developed by professional sound engineers and now marketed under the auspices of semi-conductor mass producer Cirrus Logic. It’s had the highest number of licensees for many years, and cast the mould for routing any input to any output using Ethernet hardware and software. The Holy Grail was – and still is – remote control of every device on the network and fast routing of audio without complex analogue switching – the switching being an integral part of the network and protocol.

EtherSound became CobraNet’s first serious competitor in 2001, another Ethernet-compliant solution satisfying the IEEE 802.3 standard. It claims bi-directional, deterministic, low-latency transmission of audio using up to 64 channels at 24-bit/48kHz. The Cat5 cable also carries embedded control and monitoring data. The sampling frequencies can be increased if you use fewer channels, so it really is adaptable to the user’s requirements. Both EtherSound and CobraNet provide the opportunity at least to build networks using different brands of pro audio equipment, but in practice, as we shall see, brands of a feather flock together.

‘Dante’ is one of the latest networking protocols to hit the streets, although even more are being touted from blog to blog. This one has been invented by Audinate, an Australian start-up with roots in corporate IT and comms research. Also Ethernet-based, Dante enters the IT network infrastructure affray with both audio and video well to the fore in a true multimedia thrust, and has already picked up a prestigious technical innovation award back home.

After this it starts to get even more proprietary. There are three levels of Ethernet compatibility or ‘Layers’, ascending in order of flexibility and adaptability and culminating in one that uses Internet Protocol (Layer 3) – although it is not, ironically, optimised for distribution of audio over the internet. Layer 1 protocols are the most customised: A-Net by Aviom; AES50 or SuperMAC, now owned by Midas Klark Teknik (and therefore Behringer); M11 by AudioRail; MaGIC by Gibson; the aforementioned EtherSound by Digigram; NetCIRA by Fostex; and REAC by Roland. Layer 2 protocols use standard Ethernet packets: AES51, which enables ATM services over Ethernet including AES3 audio; AVB (Audio Video Bridging), on its way from the IEEE AVB task group and already touted by the AVnu Alliance; CobraNet; Prosys, as used by Harman brands; RAVE by QSC Audio; and Hydra by Calrec.

Layer 3 protocols use standard IP packets: AoIP by Wheatstone Corp, used mostly in broadcasting; Livewire by Axia/Telos; Dante by Audinate; and QSC Audio’s Q-LAN – developed by Rich Zwiebel, the father of CobraNet, in his new role at QSC. Away from Ethernet, there is the well established but reinvigorated Madi or AES10, developed by the Audio Engineering Society using 75Ω coaxial cable and BNC connectors. Optocore, from Germany, is the fibre-optic solution: a synchronous ring network of up to 64 audio channels with access to most other media and data formats.

Perhaps the most telling comment about the standardisation of networks comes from Zwiebel himself, who does not believe that anything has effectively superseded his own creation since those pioneering days at Peak Audio. ‘Each protocol has its own characteristics,’ he explains. ‘At QSC we’ve just launched our own solution called Q-LAN. There’s a lot of baggage that goes with co-developing with other brands. We have our own priorities: Gigabit Ethernet; the ability to run over routable networks; and being Layer 3 is essential, we feel. As our portfolio grows, everything gets included.’

‘I’ve got nothing against being open platform per se, but after you’ve spent so much time on your own solution you’re starting from scratch again to try to make it universal,’ he continues. ‘Q-LAN uses IT network standards, anyway. I’m hoping that the pro audio industry will embrace Ethernet more. It may not be ideal for live sound yet, but fixed installation is fine. With 32-bit audio, our latency is just one-third of a millisecond in and one-third of a millisecond out.’

Mission control

If the pro audio system is now a network, its hub is the mixing console – if indeed we can go on calling it that for much longer.

‘We’re gaining market share among users who recognise that the console can manage so much more than just mixing, which was impossible with analogue desks,’ says James Gordon, MD of DiGiCo. ‘At a recent installation at The National Theatre in London, we’ve seen a complete overhaul of our distribution of I/O. Before, we would have used patchbays and tie-lines, but now everything is via the consoles. Consultants have to stop thinking of a console as a console, and think of it as an audio network.’

Sheldon Radford is senior product manager for Live Sound at Digidesign in the US, and agrees that the Venue series consoles are rapidly becoming network providers as well as mixers. ‘In the US, the house of worship market is important in this trend,’ he says, ‘as well as regional theatres and nightclubs. In Europe, theatres that host tours as well as local events have similar requirements such as knowing how much the console can handle and replacing all kinds of ancillary equipment. Customers see both technical and commercial advantages. In a fixed installation you can set up scene-based automation for all the different events that happen, with reliable and consistent recall in a very small footprint. This flexibility holds true for live shows coming through, but consider also the kinds of network you can call up and control very quickly for conferences, conventions and any number of applications.’

‘Digital control is everywhere nowadays,’ points out Leon Phillips, product manager at Allen & Heath, who adds that this means ‘the market is primed for ‘smarter’ systems. Having presets for different uses is a huge advantage of digital, and when users set up recalls for each application – instead of going through manual configuration – this now extends beyond the basic sound mix to countless parameters concerning the venue, the event and the circumstances. You don’t even have to be particularly technical to reset sophisticated settings at the touch of a button, with increasing control over video, lighting and other aspects of the environment as well.’

‘The key to it is Digital Signal Processing,’ he continues. ‘The software built into the console can take care of several functions that used to be handled by equipment outside of the desk, such as delay lines, crossovers and paging interfaces. You’re also dramatically cutting down on wiring and general hardware management, because if the system is DSP-managed there is simply no need for all those balanced cables coming in and out of the console. Then take away all of the mains distribution, rackmount housing, PAT testing and maintenance, and you have huge cost savings.’

System addicts

‘From the very beginning, we wanted the XL8 to be a networked audio system rather than a digital mixer,’ says Midas brand development manager Richard Ferriday. ‘This is key to having an audio and data network which is fully integrated into the design rather than bolted on later. It also allows for a modular, failure-tolerant system and permits the incorporation of additional features other than those associated with audio. If you look at the back of an XL8 Control Centre, you’ll find more video connectors than audio XLRs.’

‘Four of the XL8’s five daylight-viewable screens have an external VGA input, so the CC can be used to monitor and control other aspects of a venue’s technical infrastructure other than audio – including video surveillance of the conductor, stage manager and others. We also have the three-way KVM [keyboard, video, mouse] switch, which provides remote monitoring and control of up to three external computers that could be running absolutely anything. We have one XL8 in a venue that uses it to run the climate control!’

‘Since Yamaha launched the PM1D, smaller scale systems have been introduced based on EtherSound or CobraNet networks,’ adds Andy Cooper, manager of Yamaha’s Commercial Audio Support Centre for Europe. ‘In each case, there are two main reasons for adopting the word ‘system’. Firstly, all necessary DSP is integrated including effects, compressors and gates and GEQs; and secondly, a network of remote input/output devices is used, removing the need for analogue multicore cabling.’

‘Obviously this means that no extra rack devices are needed for mixing or processing sound, greatly simplifying the specification and installation procedures. But some of the cabling infrastructure can be of a specialist nature, which would need to be considered during the design stage. Even so, in many cases industry standard fibre optic or Cat5 cables can be used – which may even be pre-installed in the venue, making life easier still.’

Standard fireworks

Scott Fraser is output product and audio network specialist at Yamaha Commercial Audio, and regularly uses all of the current Ethernet-based audio protocols – a telling sign of how Yamaha ensures that its consoles are compatible with most of the networks on offer. ‘Each protocol has its own advantages,’ he confirms. ‘CobraNet, EtherSound and Dante all comply with the IEEE 802.3 standard, but their compatibility with IT environments will vary. It’s essential to establish exactly how the audio packets are distributed, and to measure this performance for how it suits each particular location.’

To take one such example, a recent Yamaha installation within London’s highly reflective Westminster Abbey uses CobraNet as the protocol for its networked audio. ‘We used CobraNet chiefly for its dual loop redundancy,’ Mr Fraser explains. ‘Every CobraNet device is connected to a primary and a secondary network switch, and if something happens to the primary ring then CobraNet automatically jumps to the second one. It jumps back at the next available point, getting you past any single point failure. Then there’s Multiple Spanning Tree (MST), which locates a virtual break in the ring to prevent continuous feedback – all of which is ideal for Westminster, because we’ve been able to protect the system from up to four points of failure and still have audio passing in all locations.’

The industry awaits a dominant standard for networked audio, but it’s by no means certain that there will be one over the next few years. It seems likely that the market leaders will remain EtherSound, CobraNet and Dante, while the growing number of proprietary protocols will continue to serve the brands that they belong to without paying too much heed to the demands of global interoperability. Working Group ballots on the IEEE 802 AVB standard are currently taking place, but every single variant of the technology can shift status from ‘standard’ to ‘amendment’ to ‘revision’ at any time. Can we expect pro audio manufacturers to tow these tangled lines? This is not an industry for clipboards and compliance, so don’t hold your breath…