First DIHARD Challenge -- System Submissions and Scores

This dataset contains all submissions to the First DIHARD Speech Diarization Challenge (DIHARD I) as well as the output of the official scoring tool for these submissions. We are releasing it to the research community to support the development of new evaluation metrics for speech activity detection/diarization and system combination techniques. For more details about the evaluation, please consult the evaluation plan:

   https://zenodo.org/record/1199638

or the evaluation website:

   https://dihardchallenge.github.io/dihard1/index.html

**NOTE** that this release only includes the system RTTMs that teams submitted to the DIHARD I scoring server. It does **NOT** include the DIHARD I evaluation set reference RTTMs and UEMs, which are needed to score the system outputs. These reference RTTM/UEM files are distributed by the Linguistic Data Consortium (LDC) as part of LDC2019S12 and LDC2019S13:

• LDC2019S12  --  First DIHARD Challenge Evaluation - Nine Sources
• LDC2019S13  --  First DIHARD Challenge Evaluation - SEEDLingS