"Secret" sound archive revealed

27 December 2001

Holiday special

For some innocent online fun this Christmas, try eavesdropping on the semi-secret pages of the UK’s National Sound Archive web site. They are part of a largely unpublicised trial, hidden from casual prying eyes by the lack of signposts on the British Library’s host site.

Paper publishers automatically deposit a copy of each new book with the BL, but the sound archive relies on voluntary donations. Even so, the vaults hold 2.5 million CDs, vinyl LPs, shellac 78s, wax cylinders and tapes, with the largest collection of wildlife recordings in the world.

Anyone can use the NSA’s site to search for a recording and make an appointment to go to London and listen to it. The trial is testing technology which could make NSA recordings available over the internet.

All wax cylinders have already been digitised to the CD standard (44.1KHz sampling, 16 bit coding) and copied to blank CD. Vinyl LPs and shellac 78s are progressively being transferred to CD.

“Warts and all”

No compression during storage is used because once data has been thrown away it can never be recaptured. Every hour of audio thus needs 650 Megabytes of storage space.

For online access the data must be stored on computer hard discs. RealAudio compression is used to let mono sound “stream” down an ordinary phone line in real time at 16 kbps.

The NSA adopts what Director Crispin Jewitt calls a “warts and all” approach, using no electronic cleaning to reduce low level noise on old recordings. This, says Technical Manager Peter Copeland, creates special problems; the compression system mistakes the low level noise for wanted sound, boosts it to make it more audible and then wastes bits faithfully reproducing it.

But the biggest problem for the NSA is not technical; it is finding interesting material that is in the public domain, with no copyright. So far the trial includes such diverse material as the sound of a black grouse and Margaret Thatcher.