±«Óãtv

On Monday we welcomed a new engineer to our team. Dan Nuttall has already made a great start immediately applying his up-to-date knowledge of Rails 3 and CoffeeScript to the news-linking prototype work. Olivier spent time with the news-linking team planning our final sprint and preparing for the final presentation of our past 3 months of research and prototyping. I started looking at methods for automatically discovering what the newspapers are talking about, and Duncan looked at incorporating interesting external links from our correspondents social-network feeds into the prototype.

We have a few trials ongoing at the moment. Kat was looking after the ongoing RadioTAG trial and attended an in-house study to see how one of the prototype radios fits into the life of an avid Radio 1 fan. The Future Internet team (Pete, Joanne, Barbara, Vicky and Liz) hosted an initial user testing session for the first phase of our device-to-device prototypes, and spent some time collating and making sense of the results.

Yves made some progress on the automatic tagging of speech audio, concentrating on the World Service programme archive which we have access to. He is able to automatically find DBpedia URIs that describe the content of the programme using a combination of speech recognition and graph distance algorithms. The next step is to evaluate those automatically extracted tags against tags that were manually applied by editors.

On Wednesday, Matthew and Yves attended the 10 year anniversary of the Centre for Digital Music at Queen Mary University. The day was split between talks, demos, and poster sessions. The sessions gave a very interesting insight into the state of current music informatics research at the Centre - covering areas such as segmentation, classification, auditory interfaces, sparse representations, source separation, music ontologies and linked data. Yves also gave a short talk on what he has done at the ±«Óãtv since he left the Centre.

Duncan blogged about a small application he's released which lets you create your own ±«Óãtv QRCodes.

Tristan, Yves and I have been thinking about how to increase the formal research output from our team. Tristan found a list of published outputs from ±«Óãtv R&D, while Yves collected a list of potential conferences and journals.

On Thursday George went to the morning session of the . He saw some "cracking new thinking" and learnt a lot about how to make innovative radio that mixes broadcast with exciting new technology. While Tristan stayed behind to learn about the difference between libel and slander, how to make an audio-only game and start his collection of radio presenter cigarette cards, George hurried back to the office for 3 telcos in a row.

To round up the week Kat attended the sprint review of the Radio product team, who, together with A&M have recently launched betas of the new Radio 1 homepage and /radio. She discussed RadioTAG and RadioVIS to see how they might fit with this new offer.

Interesting Links

• Fresh from the EBU stand, Chris N was impressed with a demo from RadioDNS, showing how to between broadcast and IP seamlessly.
• of a new study on online behaviour and TV consumption in the USA. Olivier says "this is very much a confirmation that we need to stop using the terms 'second screen' when it is increasingly obvious that the living-room TV is not always the 'first screen'".
• We enjoyed James Darling's over on the BERG London blog on creating applications using the TFL bus countdown data. Especially the bit where he taped a Kindle to the office door.