Welcome to the DyViS Project
Click here for a recent report on research from the DyViS project
ESRC award no. RES-000-23-1248
October 2005 - September 2009
More and more court cases involve the need to establish the speaker of some recorded speech - a hoax emergency call, a fraudulent phone transaction, an obscene voicemail, and so on. Voices, however, are not like fingerprints. A person's voice varies, depending for instance on tiredness, how loud and fast he or she is speaking, and many other factors. Despite this, there is a core of similiarity in an individual's speech. This project will record 100 speakers of the same accent in different speaking styles, and analyse their speech to determine how far they can be discriminated and what the best measures are for characterising their speech. The effects of using the telephone on an individual's speech will also be analysed.
In particular it will explore two ideas. The first is that speakers' 'vocal signatures' lie in the rapid, transitional movements of the speech organs between sounds. The second idea is that within a homogeneous speech community the sounds which are most likely to differ between speakers are those which are undergoing a rapid change in pronunciation over time. The results will show which parts of the speech signal forensic speaker identification should focus on. The project will also provide a carefully controlled large-scale speech database for further research.
The project is funded by the UK Economic and Social Research Council. We are also grateful to BT for sponsorship relating to the telephone transmission aspect of the investigation.
Thank you to all volunteers who participated in our recordings. The database is now complete. Click here for details about the database, including how to order a copy.