Since 2009, when I first saw AVB at Infocomm, I've been writing here about audio networks and their applications for live sound. Over that seven years, the audio networking landscape has changed pretty dramatically. I was a big booster of AVB, but in the end AVB just didn't get much traction in the live performance audio market, and four years later I wrote, "Audinate's Dante has won the race, while AVB is still being tweaked by the mechanics over in the pit lane." And while AVB (also now known as Time Sensitive Networking (TSN)) is a great technology, as I wrote after Infocomm last year, Audinate's Dante is now really the dominant player in our live sound market. So this year at Infocomm it was pretty much more of the same, but there were some interesting developments.
Audinate's Dante Audio Networking World
On the day before the Infocomm floor opened, Audinate held their annual Dante AV Networking World event. The highlight of this event for me in recent years has been Roland Hemming's great talk on the state of the audio networking world. Here's one of the key slides from his talk, showing the number of manufacturers who have licensed each protocol.
And this is the count of products:
Some of Mr. Hemming's research is funded by Audinate, but from what I can tell his methodology is sound, and he is addressing the audio market that I live in--excluding automotive, consumer and industrial industries, which are certainly future growth areas for AVB/TSN (and in the other fields I cover, industrial control for shows, this is certainly a promising development).
The key thing that I've been saying all along is that our live sound market is driven by end users, with manufacturers making decisions like what audio networking protocol to use and then selling us a solution. To me, once Yamaha--the 800 pound gorilla of the professional live sound market--started developing their entire product line around Dante in 2012, the choice of an audio network was made for many users. And my opinion is supported by Mr. Hemming's research; here's another slide from his presentation:
My crappy cameraphone picture may be a bit hard to read, but the reasons for selecting a protocol went from "Most Products" to "Will be around" down to "Easy". And, at least for the survey respondents reached by Mr. Hemming, that has translated to a lot of Dante projects:
So the continuing success of Audinate and Dante was not hard to see at Infocomm 2016; Dante was everywhere on the floor, and they had a lot of product in their booth (I actually got distracted there and forgot to take photos of the whole booth):
And one of the cooler Dante things I saw at the show was Amphenol's new Amphe-Dante endpoints which cost just a few hundred dollars:
Dante is a proprietary standard, and I of course would always prefer open standards. But seeing innovative, affordable products like this Amphe-Dante interface coming out is yet another sign that Dante is effectively a de facto standard in the live sound market.
Another sign of the maturation of Dante is that at this show they rolled out two levels of Dante certification; a third, more advanced level is on the way. Level 1 is an online test that is pretty straightforward and easy to answer for anyone who has worked with Dante. Level 2 involves an online test and a practical test which right now is administered by Audinate at these kind of events. I've been working with this stuff for a while so I actually skipped out of the Level 2 training and jumped right into the practical test (avoiding a huge line later), but I passed, and can now display the logo to the right.
I think certification is a really interesting development and a smart strategic move for Audinate: further positioning Dante as a standard, establishing a baseline of knowledge, and also (of course) benefiting on the marketing front.
Another interesting development that was reported at the presentation that Audinate has left the AVNU alliance (AVB trade association), although some friends pointed out that some of the biggest Dante licensees (Yamaha for example) are still members. This slide from the Audinate presentation sums it up:
And that leads to the next thing, with Dante so widespread in the live sound market, what about the companies that build their product lines around other solutions?
An Audio Ecosystem--Developments in Network Connections
In my professional work in live sound system design and alignment, I'm a big Meyer Sound fan and user. And Meyer made a big commitment early on to AVB, and their products are built around it I used this picture in last year's blog to show why I hoped Meyer would at least include hooks into the Dante world; that's Yamaha Dante-based RiO boxes on the left coming out of the console system, and Meyer Galileos on the right feeding the speakers, and a whole lot of analog XLR running between them.
Note: The picture above shows Meyer Galileos which didn't have any audio network interface, but if Meyer continued on an AVB only path, the photo is what future interconnection would look like.
Apparently I wasn't the only one wanting a networking connection from Meyer's AVB world to the larger Dante world; they had on display the replacement for the rightfully beloved Galileos (which have a limited future because of declining parts availability): the Galileo Galaxy line of processors.
Note the Extreme networks switch on the bottom of the rack. It was until recently (see below) the only AVNU certified switch, with the AVB option costing about an extra $400/switch.
Friends on the Meyer booth told me that this Galaxy system--in addition to native AVB support--will eventually offer either a Dante interface or AES-67 open standards based audio inter-operability (see my writeup on AES-67 as an interoperability interface from 2013). So for the future, the mess of XLR cables I show in the picture above should eventually be able to be replaced by a couple Cat 5 cables.
How this plays out will be interesting to see, since Audinate has included AES-67 support in their products for a while; it's now up the individual licensees to implement it. This situation was presented by Audinate's Brad Price in a series of slides at the Audinate presentation. AES67 has been included since April 2015:
They are currently in the manufacturer testing phase:
But there may be some complications:
This discovery issue is an interesting issue, but I think even if it has to be manually configured at first there is enough market push to make this happen.
And while Meyer talking Dante compatibility was a new development, as in past years other natively AVB-based live sound players were showing Dante interfacing--Avid had this on their stand:
And Digico, who are heavy MADI users was showing this again:
So it's pretty evident that the majority of players in the live sound market are embracing--in one way or another--Dante. But what about AVB?
So is AVB dead? Absolutely not. Dante is dominating the live sound market, but AVB/TSN is a well engineered solution that has found a home in portions of the live sound market (most notably Meyer Sound, Avid, Pivitech and MOTU). And at the AVNU booth I had a fascinating conversation with my friend Greg Schlechter who is a technology marketing strategist for Intel and Graeme Harrison of Biamp; they view the whole world of audio networking as an ecosystem with Dante and AVB and Ravenna all as parts of that ecosystem.
Biamp is probably the largest manufacturer of AVB product in the A/V market, and at the show they announced the ability to send video over AVB (I think this is the first company to do this in released, end-user oriented products), which is very cool. And as Graeme explained Biamp's decision to use AVB as the core of their product line it seemed like a very sound decision to me: they wanted to be freed from a single supplier who could disappear or change direction, the way Cobranet did when it was sold to a larger chip manufacturer who eventually stopped development. But Biamp occupies a slightly different market space than the world of live sound where I operate--they primarily provide complete, integrated, "turn key" solutions to customers like airports or hospitals. Of course they have to do some system integration, but their main customers aren't often patching up a bunch of mics from manufacturer A into a mixer from manufacturer B to drive speakers from manufacturer C, which is what we do in the performance audio world on a daily basis.
All that said, AVB/TSN is finding acceptance in the automotive and industrial markets, and one of the big developments on the AVB front at Infocomm is that Cisco, the dominant player in the networking hardware market, now (finally) has AVB support in some of its higher end switches:
This is a great development, and had this switch been 7 years ago, the live sound market might be dominated by AVB now.
Update June 27: Arthur Skudra on the Theatre Sound Mailing List and Kevin Gross pointed out that Coveloz were demo'ing AVB AES-67 connectivity, demo video here.
Ravenna had a booth at the show, and it does seem that this technology is gaining wider use in the broadcast world, but hasn't crossed over much into the live sound arena. Having different systems operating between live and broadcast audio isn't much of a big deal since Live and Broadcast are pretty much separate disciplines, but once AES-67 interoperability fully comes to Dante, this will give us an easy connection to the broadcast world.
I should mention too that the Open Control Alliance standard effort, spearheaded by my old friend Jeff Berryman of Bosch, led to an AES standard of AES-70, and unfortunately the picture I took is way too blurry to include here. The most interesting development I saw on the stand was that the industrial control manufacturer Beckhoff, whose gear has now found widespread use in theme parks and in stage automation has continued their push to embrace our market by incorporating AES-70 into their products.
I storm chased my way 8000 miles to and from Infocomm this year (photos here and of the amazing show Meow Wolf in Sante Fe here), and therefore had a lot of time to think. And one thing that I've been thinking about lately (which I hope to write up further later) is the way our market as a whole has matured in terms of business, techniques, and technology. It occurred to me going through an old missile silo in South Dakota on a boring weather day that things seem old and get obsolete when their underlying technologies are no longer modern. For example, this missile silo was built with almost all analog stuff, and while it was amazingly engineered, it just seemed old, because it's just not the way we would build things any more.
But even though the newer generation of missile control systems--which is now pretty old--uses obsolete 8" floppies, it's still a computer so it's not really all that different fundamentally than what we would build today. With networks, the underlying technology--Ethernet and IP--is likely to be useful and current for the rest of my career. Maybe we move to fiber, maybe we get a bit faster, maybe we go to IPv6, but I would be willing to place a sizable bet that 25 years from now we'll still be using IP and copper and fiber and radio.
And that means now that we're on the network as an industry (and nearly every department uses one now), the playing field has changed, and the constant acceleration of dramatic change we saw (for example in the 80's with the initial transition to digital audio), may have slowed a bit. And that's a good thing for the industry. For example, at my school I just got us a Yamaha CL-5 mixing console. I got that console because it is widely used in the audio market that I see here in NYC where my graduates work, and we want to train our students on commonly used stuff. But what's interesting is that the vintage 2004 Yamaha PM5D console we replaced is still fine (we will use it for monitors)--the only real difference from the user perspective of the new console is native Dante and a touch screen. And the CL-5 will probably be in our inventory for at least 10-15 years (our speakers are now nearly 20 years old and still working fine), and that means we are using Dante for the next 10-15 years. After that? Who knows. But with the market penetration of Dante now, any future solution will have to accommodate it. And if something new comes along, it too will likely use IP and Ethernet. And audio will still be audio, so as long as we have something like AES-67, we'll be good.
So my advice continues to be don't choose an audio network, choose the products you want and then figure out how they can connect. As long as you build a workable network (and I have a book about how to do that) you'll be fine.
UPDATE July 13, 2016: Comments from Graeme Harrison of Biamp:
I thought that your blog article was well-balanced for the live sector of the market (things are very different in the install sector) and well written. A couple of things that I would pick up on are:
1. I wouldn’t consider the Cisco Catalyst 3650 mini switches to be that high end - about $1,800 (with the correct level of software to support AVB) for a 24-port switch with fiber uplink ports. Another point is that Cisco are including AVB in their core software - it isn’t an extra cost like Extreme (and Netgear). This is the way that we want to see things going with the ability to run deterministic data just being a tick box in a broad list of switch specifications.
2. The point about ‘AVB AES67 connectivity’ is a little obscure. What was actually happening was that Coveloz were showing AES67 (Ravenna is this case) running over AVB. What this means is using the 802.1 base level protocols already build into the AVB (or TSN!) standard is provide guaranteed transmission rather than ‘best effort’ data flow. Obviously, if one is doing this (and therefore using AVB enabled switches in order to accomplish this) one might as well just use the whole AVB standard to keep things simple (as well as to reduce the cost as there are then no top level license fees to be paid or proprietary hardware to be bought). I appreciate that in the live sound environment, probably best effort works just fine with the correct manual switch configuration, because the network is separated and is reserved for audio. We see in the future more convergence of video with the audio (indeed we, as you mentioned, introduced product to do this). Being able to guarantee end-to-end video latency is huge for our customers and this is just not currently possible without AVB. Looking further ahead (again in the installed AV market) we see convergence of deterministic data (in this case media) within data networks and for this to happen, AVB-TSN is essential.
Thanks to Mr. Harrison for the clarification!