There’s lots of buzz around ‘Visualising Radio’ here on the 7th floor at FM&T Audio & Music towers at the moment. First, there’s DAB Slideshow, then there was Scott Mill’s ‘Radio 1 on Three’ TV show (not strictly 'visualising radio' - this is just radio-on-tv!), and finally Yasser’s post right here on Radio Labs pulled a lot of it together and got people thinking.
As I’ve said before, we like to try new things here, and nothing makes us itch more than seeing a really cool idea just sit there as a mock-up an not get any further. The one which really caught our eye was the IPTV visualisation demo. It looks great, and everyone who sees it says ‘wow, that’s cool’. So what’s stopping us from turning it into a working prototype rather than something made in Adobe AfterEffects? It turns out, not a lot. So we had a go – once again, as a 10% time project.
The idea is to take the basic now-playing data from our music radio networks, throw it at the web, and see what we can get back. We could then use this, and other BBC API’s to create a pretty rich visualisation console pretty much automatically. We had a quick brainstorm and decided that we’d use the excellent Last.fm, the incredible MusicBrainz, and the usual suspects Flickr, YouTube, and LyricsFly.
Now, before we let you take a look, some caveats….
• This is a functional data demo – there’s been no visual treatment at all. In fact, it looks pretty pants
• It can be a bit buggy
• Its mainly client-side code so you’ll need a proper browser
• We don’t use the VCS playout system in the studios overnight or some evenings, so you may not see now-playing data all the time
• There’s still some work to do to optimise the results (When we play ‘Oasis’ you can guess what kind of images we get back from Flickr….)
So given that, turn on the radio and have a play…..
Not bad for a functional prototype we think. We’ve had it running here in the office for days now and I’ve glanced at it more and more – and watched more YouTube music videos than ever before. I’ve even found out about gigs in London by my favourite bands which I didn’t know about, just by glancing at it every now and again. Thanks to BBC Backstage for the hosting, support and general cheering on. They’re great guys.
Right, All done? Hardly. So, what next?
Well, what we want to build is a full-screenable, possibly Flash-based visualisation console. The idea being you can embed it in a webpage where it’ll work in a nice little window. But that there’s a ‘full screen’ button where (much like iPlayer) clicking it gives a larger rendered experience. We’re some way along the line of doing this now using exactly the same feeds we use in the DHTML version. We've got some rough visual designs for this and they're progressing really fast - once we've got some more stuff you show you, we'll of course post it here. But this is kinda where we're going...
Eventually, we’d like to incorporate live video streams from the studio where the web’s metadata about the current song enriches the otherwise fairly dry experience of watching a DJ speak into a mic – but we need a bit more infrastructure for that.
We’re also working on the technical infrastructure behind it. At the moment, it’s a little shonky, with the app polling an intelligently cached xml file which contains the now-playing data. Ideally we’d like to move to something like Jabber or Comet for efficient asynchronous real-time data transfer but we need to prove the interface and audience appetite first.
So why’s this any good? Well as we learned from Yasser, Visualising Radio is all about ‘glanceability’. Its also about automation – we couldn’t do this if it was human-intensive. So this prototype shows its possible to get relevant and useful information automatically, just by knowing the artist and title of the current song, and that it really does enrich the listening experience – its incredible the amount of YouTube music videos we’ve seen out of the corner of our eyes in the last month or so.
What we need to do now is make it visually stunning (a-la Yasser’s Colin Murray IPTV prototype) and work on the data to make a truly compelling offering.
Before I go, just a quick nod to the other people who are doing similar things with either our data or their own.
• Chris Riley is taking our feeds and helping you buy that music on Amazon, and pulling in similar data to us from Last.fm.
• FoxyTunes have got a similar idea of taking an artist, throwing it at the web and seeing what comes back – often some really cool stuff.
• SleeveNotez do a mashup with your Last.fm now-playing data and again crawl the web for useful nuggets of data to enrich your listening experience.
• Finally Minty takes the slightly different approach of creating a psudo music TV channel based on the live Radio 1, Radio 2 or 6 Music now-playing data. The idea is that you can watch this independently of listening to the radio, and you’ll get all the music Radio 1, say, are playing but with added pictures.