This
drive contains all of the anonymized (muted) sound files from the Audio
part of the British National Corpus, together with the associated
time-aligned transcription files. These are in the form of Praat
TextGrids. Each TextGrid has four tiers (layers of description): 1)
phonemes (labels and their start and end times), 2) words (and their
timings), 3) muted words (i.e. anonymized portions) and their timings,
4) "errors" - the location of various kinds of suspicious alignment
errors; this latter is a kind of quality-control data. All of these
files are suitable for public release according to the original
agreements with the participants. The copyright and access terms
governing (a) the recordings and (b) the time-aligned transcription
files we have created are spelled out at
http://www.phon.ox.ac.uk/AudioBNC.