AMI corpus download
Use this page to download signals and annotations from the AMI corpus. The annotations, which include the orthographic transcription, come all together in two zip files: one for manual annotations and one containing automatically derived data. The signals are too large to package in this way, so you need to use the chooser to indicate which ones you wish to download. Full-size videos are not available from this page; for them, you need to order a fire-wire drive distribution of the data using the contact page. See documentation for more information about the annotations and signals.
Annotations, including transcription
Annotations are in NXT format. To use with signals downloaded below, unzip one or both of these files into the 'amicorpus' directory. Requires NXT version 1.4.4.
- AMI manual annotations v1.6.2 10-Apr-2017 (22MB) - annotations unchanged since 16-June 2014 release; license altered to CC BY 4.0
- AMI automatic annotations v1.5.1 10-Apr-2017 (68MB) annotations unchanged since 09-Aug-2011 release; license altered to CC BY 4.0
- DOME (DOminance in MEetings dataset) annotations and dataset (5K CSV). See documentation.
- Social role annotation (0.7MB compressed CSV. See Sapru & Bourlard 2015).
1) Select one or more AMI meetings
NOTE: For scenario meetings, 1 day-recording session is divided into four [a, b, c, d] 1-hour meetings. Selecting ES2008 meeting session together with 'a' below allows you to get signals for ES2008a meeting.
All of the signals and transcription, and some of the annotations, have been released publicly under the Creative Commons Attribution 4.0 International Licence (CC BY 4.0).