The collection experiment was designed and executed by a highly experienced team of data acquisition experts and signal/image processors.

More than 300 people with substantial diversity and equal mix of genders

Multiple sessions (4) for each subject spanning two to six months

Face Data

- Color, high resolution stills (1024x768) and DV video
- Differing facial expressions
- Wide variety of angles, including fixed and swept
- Lighting changes, severe and subtle
- Blue screen Background
- Additional “Free form” uncontrolled images available

Speech Data

- Industry standard, high quality 44.1kHz/16 bit sampling
- Professional microphone and telephone handset recordings
- Multiple utterance types (words, number, sentences, etc)
- Multiple repetitions of individual utterances
- Twelve “token sets” of utterance types

To hear the speech samples, Press the play button below each subject's photo.

Multimodal

- Continuous digital video for all subjects/sessions
- Additional continuous digital audio for all subject/sessions

Accompanying description data files in XML or ASCII format

This database is not operated as part of a competition, and is available to be licensed in full or in part by commercial, government, or academic entities. Pricing is dependent on the subset of the database requested. Licensing agreement is required.

If you are interested in obtaining the database, or in downloading a single subject sample, send an email or call 734-994-0028.