Skip to content

Latest commit

 

History

History
42 lines (30 loc) · 1.3 KB

CONTRIBUTING.md

File metadata and controls

42 lines (30 loc) · 1.3 KB
Raw

Contributing

Your contributions are always welcome!

Just send a pull request, and I will review and merge it.

Guidance

General

For new items, please follow the existing format.

Publications

Only include complete papers that are directly related to speaker diarization.

For example, these will NOT be accepted:

  • Course project reports.
  • One pager description of a diarization system submitted to DIHARD challenge.
  • Commercial system technical document.
  • Media post.
  • Low quality paper without experiments and evaluations.
  • Publications not directly related to speaker diarization:
    • Pure ML papers.
    • Speaker recognition papers.

Software

  • A Framework is a software that has all the necessary features to perform speaker diarization, including audio processing, feature extraction, speaker analysis and clustering, etc.
  • A Evaluation software must be able to produce speaker diarization related metrics that are permutation invariant, such as Diarization Error Rate (DER).
  • A Clustering software must correspond to a clustering algorithm that has been used by at least one diarization publication.

Dataset

A diarization dataset must contain utterances with multiple speakers speaking in turn, and each utterance must have time-stamped speaker annotations.