Submit Challenge Session Entries Here.
The second Multi-Genre Broadcast Challenge (MGB-2) will evaluate systems for transcription and alignment of multi-genre Arabic TV broadcasts. The Challenge uses a fixed training data set of around 1,200 hours of diverse Arabic TV data from Aljazeera, covering broadcasts over a 10-year period. Transcriptions derived from lightly supervised alignment are supplied. The data is available under license to all participants in the Challenge.
MGB-2 follows a similar format to the first MGB Challenge, which focused on English TV broadcasts and was an official challenge task at ASRU 2015. Participants may enter any of two tasks: speech-to-text transcription, and alignment of broadcast audio to a subtitle file.
See the website: http://www.mgb-challenge.org/
Dialog state tracking is one of the key sub-tasks of dialog management, which defines the representation of dialog states and updates them at each moment on a given on-going conversation. To provide a common testbed for this task, the first Dialog State Tracking Challenge (DSTC) was organized. More recently, Dialog State Tracking Challenges 2 & 3 and Dialog State Tracking Challenge 4 have been successfully completed.
In this fifth edition of the Dialog State Tracking Challenge, we will continue evaluating the dialog state tracking task on human-human dialogs. Different from DSTC4, we will focus on cross-language state tracking. In addition to this main task, we also propose a series of pilot tracks for the core components in developing end-to-end dialog systems based on the same dataset. We expect these shared efforts on human dialogs will contribute to progress in developing much more human-like systems.
See the website: http://workshop.colips.org/dstc5/