 Okay. So we start again with another lightning talk, this time by EBU Ben Poe who is speaking about smart radio for smart devices. Hi, can everyone hear me at the back? No, thank you, Willem. Okay, so this is just a bit of mental sorbet between during or after lunch, very quick lightning talk, is to discuss something about smart speakers. This is something that as the radio guy for EBU that I've been looking at for quite a long time, it's something that our members are very involved with. But this is about really how smart speakers and broadcast come together. So a little bit of background, apologies for the transition. A little bit of history about smart speakers. Kind of started around 2015, we saw Amazon Alexa, Google Mini, Google Home. All these other platforms including Apple, Sonos and I think in 2018 where this timeline ends, we're releasing some sort of voice control microwave which I thought was a joke but isn't. In terms of sales, various wild projections happen but essentially we were looking at 33 million devices by the end of 2017, about the same in 2018 and massive growth in smart speakers by about 2022 and 2024. It's a huge potential market. What are they actually being used for? So there's some statistics being done on this and there were two questions asked. What reasons would you have to buy a smart speaker and what do you actually use your smart speaker for? So a couple of statistics there. We see that people buy them because they want to listen to music. They want to ask questions without needing a type and then we see what people actually do with them. So they play music, fantastic. They get the weather. They ask general questions. They control devices. They check the time. So I've highlighted a few things here, what I think are traditional things to do with radio. Radio is pretty good at this. It's pretty good for music. It's pretty good for entertainment. It gives you information. It usually gives you a fairly good sound, some might argue with that. Talksports, music, selection, these are all things that radio is strong at. So we ask the question, are these smart speakers the new radio devices? And here's the statistic which is 39% of respondents replace the AM FM radio with a smart speaker in the US. They actually get rid of their existing devices and change that for a smart speaker. Another thing I think is important as well is that some people are replacing TVs, tablets and computers with smart speakers. They want to minimise their screen time. They prefer talking to things, which is revolutionary. Now smart speakers are great. They're very useful. People buy them. They're very cheap. But there's one big problem with smart speakers in that they only do internet streaming. They don't do broadcast reception on the whole. Broadcasters still, certainly for EBU members and broadcasters throughout the world, are very important platform. It's a vital platform. The vast, vast majority of current radio listening still goes over broadcast. And for very good reasons. It has increased coverage. It's more reliable and it's way more cost efficient. We did a study last year or the year before which says that FM for instance is multiple times cheaper than IP but DAB is about six or seven times cheaper than FM is. So really it's a much cheaper platform to deliver audio. So could they do broadcast? Well we asked a few manufacturers to do this or to think about it and they do exist in different modules in a radio device but they tend to be moving between modules. So an example of that is you asked your voice control radio to go into FM or DAB mode and after that you can't control it. So you're kind of shifting from mode using voice. It's a very jarring user experience. What about a unified service list? Listeners don't care whether they're listening on FM, DAB, internet streaming. They just want to listen to a radio station. We talked to some manufacturers. They said it wasn't possible. So we decided to do something ourselves. So we have this rather attractive looking blue radio which uses voice to pick a station. You say Alexa because we chose Alexa. That was the quickest thing for us to hack very quickly. Play radio station. Play BBC Radio 1. Play Rye 1. Play Radio Pop in our example. And the radio responds back to you saying playing Radio Pop and it streams up. The clever thing I think about this radio is that you don't care about how you listen to the radio. It automatically selects itself the best way to listen to the station, either FM, DAB or IP. And it uses data that's controlled by the broadcaster. It uses openly standardised metadata which is controlled directly by the broadcaster. Now there's a video here but I've only got 10 minutes. There's a very quick blast through so I'm going to skip over that. But there is a YouTube link there if you want to see it. I haven't done a QR code because I'm not that clever. How does it work in the background? What happens is you ask your radio device to play a radio station. It then goes to the Alexa voice service. So again we've kind of hacked around with this a little bit. What Alexa voice service does by default is it goes and looks at something called tune-in which is a streaming radio aggregator. Now what tune-in does is it returns a URL. But when the device has actually switched on, what we've done in the background because our device has both the capabilities to receive broadcast and internet is we've done a band scan. That is we've looked over all the frequencies that radio stations can exist in. We've seen that there are radio stations at certain points. We've done an IP lookup in the background using something called radio DNS which means you can go from a broadcast frequency, you turn that into a DNS entry, you then look at that DNS entry on the internet and pull back some metadata and XML. And that XML data contains ways that you can listen to a station. It contains station names. It contains things like logo branding. It contains things like where the station is, the geolocation. Gives you all sorts of interesting things to build a very new, interesting radio experience on top of that. So after this band scan we've got information about each and every radio station that we can see on FM and DAB. We know it exists. We know that you can listen on different platforms. So a local station we know you can listen on DAB, FM and IP. So when TuneIn then return us this URL, we match that against an internal database. And if we see that this station has this URL, we can also see that you can listen to it on FM or DAB using broadcast. Rather than switching to the IP stream, we switch to broadcast. So one thing about this project is we did it in collaboration with a number of people. We did it in collaboration with the NAB's pilot project. And the NAB are kind of like the equivalent to the European Broadcasting Union, the US. And they're very keen to promote listening on FM for the reasons that I've said before, in that it's a better way to listen to radio services. It scales better for their members. Why do we do this? Well, smart speakers are obviously key devices for listening. They actually represent a very small proportion of radio listening currently. But automotive is an important part of radio listening. And automotive manufacturers want to put voice controls in their cars. They're doing that. They're doing that for Alexa. They're doing that for Google. They're doing it for all sorts of other things. So we wanted to demonstrate to automotive manufacturers that, hey, you can use voice control to also control the really good broadcast hardware you've got in your cars. You don't have to have an Alexa mode that you switch from a button on your dashboard. You can actually use voice intelligently to use the existing broadcast reception to do better things apart from this closed voice control ecosystem. Also, we wanted to do this because it's not new ground. As I said, it uses open standards. It uses all these acronyms here, which you can look up afterwards. These are European standards. They're used globally. It's the same information that's broadcast over DAB. This is not new for broadcasters. It's well-implemented in chipsets, manufacturers, and client devices. So just to center the coverage, this is actually, I'll just put some figures up here. We run services for our members that support this metadata, this information. It's freely, openly available. For instance, 75% in Germany, Austria, 92% in UK, France, Italy, Spain, all sorts of other countries in Europe, but also the US and Australia. This isn't like some sort of weird Brexit scenario. This is just me trying to fit everything onto the same map. So some next steps for us. By the way, this isn't actually Vapeware. We've got a device here, which I can show you the insides of if you're interested. We want to create some open-source software. We want to do hackathons of this. So we want to actually do a hackathon this summer somewhere where we can maybe make 15 of these devices and let people loose with them. Because one thing about this is we've kind of done the bare minimum. The interface of this is not great. It's pretty bog standard. It's pretty ropey, but it does work. So we want to create a community around making this more functional and more better. What we've learned from this project and we've actually done more with it. One thing I learned from this is actually get something done rather than talking about metadata standards and good ideas for radios. Actually, spending a month, which is what we did in making this, spending a month to make a system, hacking around with the Alexa SDK, doing some recompiling and re-signing and writing some Python code, sticking on a Raspberry Pi, for instance, and then putting it inside a radio was actually a much more powerful demonstrator of what could be done for radio services. So I think I'm probably mostly out of time, but what I want to do is I want to share this and let people actually loose and make in their own. So the link here for GitHub, we've actually yesterday, I think it was, made the software fully open source and it's also got hardware designs. It shows you how to buy one of these radios, hack it around with a Dremel and actually make a radio for yourself. Thank you. We have time for one question. Hi. So if I understood correctly, you're intercepting the tuning URL. Yes. Why don't you just build a scale? A what? A scale, an Alexa scale. So you can do everything and properly integrate it with Alexa. That's the way they planned it. And that's the way you'll get it certified eventually. Yeah, so that might be the next step. But the thing about the tune-in interface is we found that it's baked in very low level. It's the default way of doing things. So we didn't want to say, for instance, play BBC Radio 1 using skill name. We wanted to say play BBC Radio 1 because it's the natural action for a listener. Otherwise you have to teach them how to use a specific skill. It would be nice if this functionality was certified by Amazon or Google, for instance, that they actually could put some sort of broadcast APIs in their SDKs. That's kind of one of the aims of this project. Thank you, Ben. Thanks.