by Greg Sell, Senior Research Scientist

A team of Johns Hopkins researchers from the HLTCOE and CLSP participated in the recent DIHARD challenge. In the evaluation, teams were given audio recordings of speech from a diverse set of conditions with an unknown number of speakers, with the goal of correctly marking the times that each person is speaking.

This evaluation was designed to highlight the situations in which current diarization methods tend to struggle and given teams an opportunity to build more robust methods for the future. When the dust settled, the JHU team finished very well, with the winning score in one of the tracks and the third place score in the other track.

More information about the evaluation and the results can be found at the evaluation webpage and more information about the JHU system can be found in the recent Interspeech paper “Diarization is Hard: Some Experiences and Lessons Learned for the JHU Team in the Inaugural DIHARD Challenge” as well as on the DIHARD Challenge website.