Find Listen Label (a.k.a. Annotatable Audio)

Today sees the public launch of the Annotatable Audio prototype, now called "Find Listen Label", which my R&D team has been developing recently. Find Listen Label is basically a wiki for annotating radio programmes. It lets BBC radio listeners go online and mark-up and annotate segments of radio programmes, creating better navigation within the programme by providing segments or chapters and enhancing the findability of the programme by annotating it with descriptions and tags about the content.



Why are we doing this? It's part of our drive to make BBC radio's programmes more navigable and more findable on the internet. We have some basic information on programmes that is used for schedules but it is not usually very rich and often doesn't really describe what is in the programme in any detail. Many of our speech radio programmes on Radio 4 or Five Live contain lots of varied and interesting content and at the moment, without listening to the programme, you just wouldn't know.

We already have a nice framework for individual pages for programmes (from the BBC Programme Information Pages project) ready for extra information and in the forthcoming Radio 4 redesign we will start to see extra metadata to allow more horizontal (topic- or keyword-based) navigation. And some programmes, like the Chris Moyles enhanced podcast, already do some "chapterising". But this particular example takes the best part of a day for a producer to do and most programme teams do not have the dedicated effort to do something like this, so we wanted to try to harness the enthusiasm and knowledge of the BBC radio audience to help us. We designed Find Listen Label (Annotatable Audio) to generate metadata that enables two things, internal navigation and findability. Internal navigation is the ability to see what a programme contains and easily move around inside that programme, moving between segments and skipping segments that don't interest you. Findability is the quality of something being locatable or navigable - through search engines, links and keywords - and by adding rich metadata about the contents of a programme we increase its findability.

We're launching the prototype with All In The Mind, a Radio 4 programme "exploring the limits and potential of the mind". We chose this as the initial programme to prototype this on as it is relatively non-controversial, factual and magazine-like with natural segments. Luckily the programme team were happy to allow us to use it as a guinea pig.



The playback interface lets you listen to the programme, skip around the segments, filter by tags and read the annotations. The edit interface lets you edit an existing segment, delete an existing segment or create a new segment. The start and end times of segments are altered in the Flash interface by dragging the left or right edges of the segment. And you can edit the title, description and tags for a programme. It works like a wiki so there is one canonical version of the programme with the most recently edited set of segments and metadata. Like every good wiki there is a history page though at the moment the history page is fairly rudimentary.

To do any editing you must be signed in. We have to follow the BBC's guidelines and for any kind of user-created content we must use the BBC's Single Sign On system. But we're not going to be actively moderating the content, just keeping an eye on it to make sure nothing breaks the BBC Editorial Guidelines too much.

This is the biggest project that the current R&D team at BBC Audio & Music Interactive have worked on. Many thanks to all the people who worked on it including Tom Coates for the initial idea, Graham Beale who created the basis of the current simplified design, Sarah Challis for the lots of work on the final design and the cool little tutorial animation, Joti Brar for project managing it to launch, Lee Goddard for the back-end Perl code and Chris Bowley for all of the lovely front-end interface. The original prototype was built in the Autumn of 2005 and you can read more about the history of the project here and here. The name change came after a number of people complained that "annotatable" wasn't in the dictionary and as an attempt to make it more obvious what it does; "find", "listen" and "label". In particular we wanted to make it clear that it was a tool, something that could be applied to a programme and not a separate application. The ultimate ambition would be to have the Find Listen Label tool available for every BBC radio, and even TV, programme.

How does it work? There's a bespoke Perl back end, written by Lee, which is basically a wiki/cvs system. It's built to run on bbc.co.uk servers, which in itself is not a straightforward task. This then serves a REST-like read and write API - when we get time we will publish the location of this API and its format. The client, built by Flash wizard Chris Bowley, is a mixture of Ajax and Flash. The Flash handles the audio playback and segmentation interface and Ajax does the annotations and the rest of the page. Speaking of the audio - it is just a progressively downloaded MP3 which is not necessarily the best solution but is definitely the simplest. It is also fairly low quality to reduce loading times. I'm sure Chris will write more about the client on his blog shortly.

This is our third or fourth iteration of the tool and it's a while since we built the first internal prototype. It's been a challenge to design a suitable simple interface, to create a common understanding of what the tool should do (it can be many things to many people), to get buy-in from management and from the radio stations, and to deploy it onto the somewhat aged bbc.co.uk infrastructure. I hope you find it interesting or useful. Please go and have a play...