metadata
license: mit
This repository contains enhanced from "TS-SEP: Joint diarization and separation conditioned on estimated speaker embeddings". The system was TS-SEP + GSS + WavLM ASR and achieved a cpWER of 5.364 % on LibriCSS.
Thie files from this repository are used in an example from https://github.com/fgnt/speaker_reassignment .
@article{boeddeker2024ts,
title={TS-SEP: Joint diarization and separation conditioned on estimated speaker embeddings},
author={Boeddeker, Christoph and Subramanian, Aswin Shanmugam and Wichern, Gordon and Haeb-Umbach, Reinhold and Le Roux, Jonathan},
journal={IEEE/ACM Transactions on Audio, Speech, and Language Processing},
year={2024},
publisher={IEEE}
}