This paper introduces the third DIHARD challenge, the third in a series of speaker diarization challenges intended to improve the robustness of diarization systems to variation in recording equipment, noise conditions, and conversational domain. Speaker diarization is evaluated under two segmentation conditions (diarization from a reference speech segmentation vs. diarization from scratch) and 11 diverse domains. The domains span a range of recording conditions and interaction types, including read audiobooks, meeting speech, clinical interviews, web videos, and, for the first time, conversational telephone speech. We describe the task and metrics, challenge design, datasets, and baseline systems for speech speech activity detection and diarization.

No abstract

No abstract

A listening test is proposed in which human participants detect talker changes in two natural, multi-talker speech stimuli sets-a familiar language (English) and an unfamiliar language (Chinese). Miss rate, false-alarm rate, and response times (RT) showed a significant dependence on language familiarity. Linear regression modeling of RTs using diverse acoustic features derived from the stimuli showed recruitment of a pool of acoustic features for the talker change detection task. Further, benchmarking the same task against the state-of-the-art machine diarization system showed that the machine system achieves human parity for the familiar language but not for the unfamiliar language.

No abstract