ai-transcribe-summarize-diarization

This repository is an unmodified redistribution of pyannote/speaker-diarization-community-1, hosted so that a desktop transcription application can fetch the weights without asking its end users to create a Hugging Face account or accept a gate.

このリポジトリは pyannote/speaker-diarization-community-1 の無改変の再配布です。デスクトップアプリの利用者が Hugging Face アカウントを作らずに 話者分離モデルを取得できるようにする目的でホストしています。

Attribution / 帰属表示

• Original authors: the pyannote team (Hervé Bredin and contributors)
• Original repository: https://huggingface.co/pyannote/speaker-diarization-community-1
• License: CC-BY-4.0 (unchanged)
• Modifications: none. The weights and configuration are redistributed byte-for-byte as retrieved from the original repository. Only this model card differs; the original model card is preserved here as ORIGINAL_MODEL_CARD.md.

The original pipeline is released under CC-BY-4.0 and, per its model card, "will always remain freely accessible".

Contents

File	Purpose
config.yaml	pipeline configuration (VBx clustering, thresholds)
segmentation/pytorch_model.bin	speaker segmentation model
embedding/pytorch_model.bin	speaker embedding model
plda/plda.npz, plda/xvec_transform.npz	clustering parameters
ORIGINAL_MODEL_CARD.md	the upstream model card, kept as the license record

Requires pyannote.audio 4.x.

Citations

If you use this pipeline, please cite the original work.

Speaker segmentation model:

@inproceedings{Plaquet23,
  author={Alexis Plaquet and Hervé Bredin},
  title={{Powerset multi-class cross entropy loss for neural speaker diarization}},
  year=2023,
  booktitle={Proc. INTERSPEECH 2023},
}

Speaker embedding model:

@inproceedings{Wang2023,
  title={Wespeaker: A research and production oriented speaker embedding learning toolkit},
  author={Wang, Hongji and Liang, Chengdong and Wang, Shuai and Chen, Zhengyang and Zhang, Binbin and Xiang, Xu and Deng, Yanlei and Qian, Yanmin},
  booktitle={ICASSP 2023, IEEE International Conference on Acoustics, Speech and Signal Processing},
  pages={1--5},
  year={2023},
  organization={IEEE}
}

Speaker clustering:

@article{Landini2022,
  author={Landini, Federico and Profant, Ján and Diez, Mireia and Burget, Lukáš},
  title={{Bayesian HMM clustering of x-vector sequences (VBx) in speaker diarization}},
  journal={Computer Speech \& Language},
  year={2022},
}

Note

For the original model card, usage instructions, benchmarks and support, please refer to the upstream repository. Issues with the pipeline itself should be reported upstream, not here.