Posted by Libby Miller on
Four public prototypes from Tellybox, recording the Next Episode, out and about with Newnews, and sad farewells to Jakub (and Freebird).
This week Tellybox released four prototypes illustrating different ways that choosing what to watch on TV can be made more fun, and a blog post explaining the ideas behind them:
Choosing is hard. Making good choices is hard. People want to spend time together and watching TV together can be a pleasant social activity. One way we can help make it enjoyable is if we make choosing an important part of the design, for example by making the process fun (“Random TV”), or helping people make choices based on the time that they have (“Clock”), or explicitly acknowledging more than one person is there (“5-2-1”), or gearing the kinds of questions asked to the kinds of mental states people want to achieve rather than making them do the work of figuring out the specific programmes that might meet these states themselves (“Something Something”).
The prototypes are now available on the BBC Taster website, so please try them out and let us know what you think.
Talking with Machines
Prototyping of the Next Episode has started, ahead of user experience testing in late February. Nicky arranged for recording of the first episode with voice actors, while Ant, Henry, and Oscar have made a start on the technical implementation required for the test.
Image: Joseph Ayre and Sarah Ovens recording the pilot episode for our interactive multi-modal voice series.
Andrew has been preparing for the user testing of this pilot, which we hope to do in February with a group of 16-30 year olds.
Tristan’s been out and about doing some talks about our newnews project from last year - the London Mayor’s Office, Conde Nast and a webinar for the INMA (International News Media Association), plus following a few leads on what might happen next with the work
Chris joined the latest BBC "standards day", which brings together everyone working across different standards bodies to look at our current priorities and future direction.
Emma has launched a user survey for her Emotional Machines project to capture the range of emotions that people experience during their interactions with voice UIs.
Oscar has been preparing his user test of different pairing methods for multimodal experiences with smart speakers. As part of this, he has been speaking to the HbbTV 2 team and to industry actors already implement pairing between two or more devices.
This sprint the Disco team have been consolidating existing work as well as looking to the future and planning new exciting things for the year ahead.
Chris has completed his work on the 'Vox' adaptable text classifier and written a technote about it.
Tim's been finishing off the work of decommissioning Freebird (the infrastructure behind our Editorial Algorithms project)- but in the process has given it a new lease of life. While it's no longer being updated, we've preserved the entire archive of 40-odd million processed and annotated news articles, and thanks to Amazon S3 and Athena, we can both store them incredibly cheaply and query them easily using a SQL-like interface. Our friends in the News Audience Engagement team have already been using this to inform some work on editorial stance detection.
Along with David, Tim’s also been planning a user testing session for our first round of 'Enriching Discovery Mechanisms' prototypes. We also kicked off another round of testing of our Quote Attribution database, with the BBC's 'Reality Check' fact-checking team.
Finally, we're all very sad to have said goodbye to Jakub, who's been an integral part of the Disco team since he joined us one year ago. He's leaving to pursue some personal projects, and his creativity, enthusiasm, and sheer prototyping speed will be sorely missed. We saw him off with an incredible meal at the London Galician Centre, and we can't wait to see what he comes up with in the future, as well as hopefully welcoming him back in some form! In characteristic Jakub style, however, he managed to find the time to complete two more incredible, weird and wonderful prototypes before departing - an 'Ambient TV' discovery interface for iPlayer, and a port of some of 'Alluvial Sharawadji', our spatial audio experiment, to the Bose Frames Audio AR glasses which Henry is investigating.
The data team have been packaging and releasing a new version of Kaldi with a revised / improved lexicon (set of words it can pronounce) based on the database of pronunciations from the BBC pronunciation unit, and some other improvements and fixes.
The team have been improving the speaker identification tool they’ve built, and measuring its performance on ground-truthed BBC content to see how it performs.
Finally, they are also developing an interactive prototype of the CAT demonstrator based on real data, including figuring out the architecture of the demonstrator and planning the remaining work.
This post is part of the Internet Research and Future Services section