Vietnam-Celeb: a large-scale dataset for Vietnamese speaker recognition

Contact email: thanh.pv.ds@gmail.com

Paper URL: https://www.isca-speech.org/archive/interspeech_2023/pham23b_interspeech.html

Citation:

@inproceedings{pham23b_interspeech,
  author={Viet Thanh Pham and Xuan Thai Hoa Nguyen and Vu Hoang and Thi Thu Trang Nguyen},
  title={{Vietnam-Celeb: a large-scale dataset for Vietnamese speaker recognition}},
  year=2023,
  booktitle={Proc. INTERSPEECH 2023},
  pages={1918--1922},
  doi={10.21437/Interspeech.2023-1989}
}

• This is the repository for the anonymous submission of the Vietnam-Celeb dataset at Interspeech 2023.

• This repository includes four parts of the dataset:

  • Part 0: https://drive.google.com/file/d/1pMuT3DFzSwib7SVcRS8VkDwPuLTsemSG/view?usp=share_link
  • Part 1: https://drive.google.com/file/d/1xayHt2HRqE1aJ4HvtUT40_9XlgvfDfRY/view?usp=share_link
  • Part 2: https://drive.google.com/file/d/1MIlM78EbN_J9cApkNes_2_BrFrf8XwMc/view?usp=share_link
  • Part 3: https://drive.google.com/file/d/1h6Na58DC03p-502B9QpC5Z_FadUwAdNA/view?usp=share_link

To extract the 4 parts, run the two following codes:

zip -F vietnam-celeb-part.zip --out full-dataset.zip
unzip full-dataset.zip

• The data folder contains the utterances of every speakers in the dataset, in which each speaker has a folder with its name being the ID of that speaker.

• There are three text files corresponding to the datasets that we have split, as discussed in the anonymous submission to Interspeech 2023:

  • vietnam-celeb-t.txt: list of utterances in the training set of Vietnam-Celeb
  • vietnam-celeb-e.txt: Pairs of utterances in the Vietnam-Celeb-E test set.
  • vietnam-celeb-h.txt: Pairs of utterances in the Vietnam-Celeb-H test set.

• We also include a TSV file containing the information of every speaker in the dataset, which include the following attributes:

  • speaker_id: ID of the speaker
  • gender: gender of the speaker
  • dialect: Vietnamese dialect of the speaker
  • source: the crawling source of the utterances of a speaker.