000933 Digital Radio Archive Speech Recognition

Description

The Library is recognised as the home of the nation’s radio archive. The current radio collection comprises around 250,000 hours. Additionally, via a longstanding arrangement with the BBC the Library also provides the only UK point of research access to the extensive radio collections of the BBC Archives. However, of the estimated 3m hours of radio broadcast in the UK each year from c.700 stations, the Library is capturing a mere 15,000 hours (mostly news-based content). It is estimated that 92% of current UK radio is not being properly preserved, with only 2% being made available for research post-transmission. To address this situation we plan to create a digital radio archive, based primarily on off-air recording, which will preserve a proportion of ongoing UK radio output (emphasising speech-based content), and make this available for research. The first step is to build a pilot radio archive over a period of two years (2018-2020), covering up to 50 stations from across the UK, to be developed thereafter into a full service. The pilot will test technologies, scalability, sustainability and research use and develop integration with British Library systems. A Digital Radio Archive Management system was the subject of an earlier tender; this tender is for a complementary speech recognition facility, with the aim of making the content of the National Radio Archive as fully searchable by word and theme as possible.