When new technologies emerge there are recognised stages to the development and uptake of those technologies. They are adoption, standardisation and convergence. Telephone systems are a good example. Photos of city streets in the 1890s with many competing phone networks, each with their own telegraph poles carrying their own cables all down the same street look silly today but they perfectly illustrate the first phase of adoption before standardisation and convergence allowed phone networks to use each other's infrastructure and become interoperable with each other. We are at a similar point in the development of audio over ethernet and more specifically audio over IP (the two are related but not the same). For the past fifteen years competing technologies have offered methods for distributing professional audio over computer networks, with all the associated advantages in economies of scale and being able to repurpose the immense R&D efforts made by the global IT industry into things useful to our, relatively tiny, pro audio industry. For an overview of the current crop of competing systems available take a look at my two part primer from last year ( Part 1 and Part 2 ) In part 2 of that primer I mention that AES67 has recently been ratified by the AES and is one to watch for the future. The broadcast industry is certainly paying close attention to AES67 and its promise of interoperability between competing audio over IP networks has had some herald it as a new MIDI. Considering MIDI is now over 30 years old and is just as ubiquitous as ever certainly illustrates the value of standards which allow interoperability. If you're not familiar with terms like network switch, the boxes which manage the traffic on a network, then there is lots of information on the basics of networking available online and to get a meaningful understanding of audio networking it is necessary to understand basic networking first. So why the fuss about this particular development? What is AES67 and what can it do?
What Is AES67?
- It is an interoperability standard
- It is for audio transport only
- It isn't a complete system. AES 67 is a feature or option in a wider audio system which can fulfil other tasks such as routing, monitoring, discovery or system control.
Why All The Fuss About AES67?
It is deployable - It is very limited in its scope. It does audio transport and that's all, by audio transport I mean moving audio around a network, I'm not talking about tape machine style transport controls. As a result it is easier to get adopted by third parties. More importantly it is Layer 3 and compatible with any modern gigabit switch and therefore doesn't present as many issues with existing IT infrastructure. The tight focus on the "network" part of network audio, using existing hardware has meant that it has come to market quickly and in an area which is developing as quickly as this that is a big advantage.
There are many things which make AES67 exciting. One of the most significant is that it is very limited in scope. The two technologies which have received the most attention on the blog are Dante, as implemented by Focusrite in their RedNet systems, and AVB which Avid are using in their S3L live sound system. The biggest differences between these two systems are that Dante is proprietary and operates on layer 3 of the OSI 7 layer model. AVB is an open set of standards and it operates on Layer 2. While AVB is a fantastic technology, it is broad in scope and its future definitely lies in more areas than just pro audio. The biggest limitation on its uptake in pro audio is probably the relative indifference end users have to which technology they use combined with the fact that Ethernet switches used on an AVB network have to have specific AVB features, ruling out old or non-AVB switches. AES67 is an interoperable stream with moderate latency, each data packet contains roughly a millisecond of audio and the total latency is 6ms, though lower and higher latencies are available as an option. It supports both multicasting and unicasting - multicasting makes all streams available everywhere but requires good quality switches. Unicasting is useful for point to point scenarios over distance where the quality of switches can't be guaranteed. AES67 speakers are coming, the Genelec 4020A prototype has AES67/Ravenna inputs eliminating the need for any extra network node/AD-DA hardware. Just plug the RJ45 into the back of the speaker - install designers will love that! Just as important as what AES67 does is what it doesn't do. It offers no routing and control protocols, no online metering or remote control and no easy web GUI management.
What About AVB?
AVB is layer 2 and to run successfully needs switches which specifically support AVB. Switch support remains an issue, its taken a long time for AVB compliant switches to come to market, they are more expensive and Cisco still don't support AVB, which is odd as they are founding members of the AVNU alliance who look after AVB development. Because of this, AVB has not been the most attractive choice for users seeking to retrofit AVB into buildings with existing IT infrastructure. That being said, AVB is very open and very broad - it is audio and video. The development and implementation of these technologies is largely driven by video and broadcast for whom the advantages are even more compelling. The bandwidth demands of video, particularly the move to 4K is driving upgrades to infrastructure and the limitation of incompatibility with legacy hardware isn't such an issue in these markets. Its applications are far wider than just pro audio and video though. Important markets include automotive and industrial control uses and the development of TSN (Time Sensitive Networks) by the AVNU alliance reflect this interest in critical control systems in non AV applications.
What About Dante?
Dante is Layer 3 and as such doesn't have the same switch compatibility constraints as AVB. AES67 support is announced and a firmware update will allow Audinate Transport Protocol and AES67 Transport protocols to coexist on the same network. The reason why you might want to do this is because AES67 is a lowest common denominator between networks and using Dante's native transport protocol might provide performance improvements when moving data around a purely Dante network. When sending audio between mixed AES67 compatible networks an AES67 stream can be used, sacrificing a little latency for improved flexibility. Dante publishes the availability of AES67 streams on the network so they can be used by 3rd party network technologies with AES67 providing the transport and Dante looking after the system control.
History is littered with examples of competing technologies fighting to become the recognised standard. MIDI is one of the rare cases where a standard is designed through collaboration across an industry and it goes on to change the world for the better. The problem with standards is that competing companies all want a standard, they just all want a standard which is very like their product! Open standards seem the best solution but their very openness can make them slow and disorganised in their development. Many of the standards we recognise today are actually licensed technologies. I remember conversations with open source enthusiasts frustrated at the uptake of mp3 and AAC over Ogg Vorbis. An interesting example came to light this week when Sony announced the end of production of Betamax cassettes. VHS vs Betamax is surely the best known example of rival formats fighting for acceptance as the standard. Everyone knows that Betamax was technically superior. What I didn't know until this week was that VHS was open and Betamax was proprietary.
The thing which excites me, and clearly a lot of the audio industry, about AES67 is that it is pragmatic. It's not seeking to win the war by defeating the competition, it's attempting to bring the competition together by letting them coexist as long as they support it. A kind of audio networking Babel fish (possibly a bad example as Douglas Adams fans will know the Babel fish started many wars!). However, at the moment rival companies are seeing AES67 as something they need to support rather than bury and because of this it seems to be gaining acceptance.