README.md

metadata

dataset_info:
  features:
    - name: audio
      dtype: audio
    - name: translation
      dtype: string
  splits:
    - name: train
      num_bytes: 885158053.092
      num_examples: 7478
    - name: dev
      num_bytes: 330788279.48
      num_examples: 1120
    - name: test
      num_bytes: 44484347
      num_examples: 347
  download_size: 1105429263
  dataset_size: 1260430679.572
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: dev
        path: data/dev-*
      - split: test
        path: data/test-*
license: cc-by-nc-sa-4.0
task_categories:
  - automatic-speech-recognition
  - text-to-speech
  - translation
language:
  - ga
  - en
size_categories:
  - 1K<n<10K

Dataset details

IWSLT 2023/2024 dataset for Irish-to-English speech translation, as originally found on this GitHub repository. The audio data consists of the Common Voice and Living Audio (IDLAK) datasets.

Data available since: Irish-English Speech Translation Shared Task@IWSLT-2023 and 2024
License: CC BY-NC-SA 4.0
Includes text/audio: yes
Shared Task Organisers 2023: Ojha, Atul Kr.; Judge, John; McCrae, John P.
Shared Task Organisers 2024: Ojha, Atul Kr.; McCrae, John P.
Contact: [email protected], [email protected]
Contributor/©holder: Insight Centre for Data Analytics, Data Science Institute, University of Galway, Ireland and ADAPT Centre, Ireland

How to load the dataset

from datasets import load_dataset, DatasetDict

iwslt2023_gaen = DatasetDict()

iwslt2023_gaen["train"] = load_dataset("ymoslem/IWSLT2023-GA-EN",
                                       split="train",
                                       trust_remote_code=True,
                                      )
iwslt2023_gaen["dev"] = load_dataset("ymoslem/IWSLT2023-GA-EN",
                                      split="dev",
                                      trust_remote_code=True,
                                     )
iwslt2023_gaen["test"] = load_dataset("ymoslem/IWSLT2023-GA-EN",
                                      split="test",
                                      trust_remote_code=True,
                                     )

Dataset statistics

DatasetDict({
    train: Dataset({
        features: ['audio', 'translation'],
        num_rows: 7478
    })
    dev: Dataset({
        features: ['audio', 'translation'],
        num_rows: 1120
    })
    test: Dataset({
        features: ['audio', 'translation'],
        num_rows: 347
    })
})

Citation

@inproceedings{agrawal-etal-2023-findings,
    title = "{FINDINGS} {OF} {THE} {IWSLT} 2023 {EVALUATION} {CAMPAIGN}",
    author = {Agarwal, Milind  and
      Agrawal, Sweta  and
      Anastasopoulos, Antonios  and
      Bentivogli, Luisa  and
      Bojar, Ond{\v{r}}ej  and
      Borg, Claudia  and
      Carpuat, Marine  and
      Cattoni, Roldano  and
      Cettolo, Mauro  and
      Chen, Mingda  and
      Chen, William  and
      Choukri, Khalid  and
      Chronopoulou, Alexandra  and
      Currey, Anna  and
      Declerck, Thierry  and
      Dong, Qianqian  and
      Duh, Kevin  and
      Est{\`e}ve, Yannick  and
      Federico, Marcello  and
      Gahbiche, Souhir  and
      Haddow, Barry  and
      Hsu, Benjamin  and
      Mon Htut, Phu  and
      Inaguma, Hirofumi  and
      Javorsk{\'y}, D{\'a}vid  and
      Judge, John  and
      Kano, Yasumasa  and
      Ko, Tom  and
      Kumar, Rishu  and
      Li, Pengwei  and
      Ma, Xutai  and
      Mathur, Prashant  and
      Matusov, Evgeny  and
      McNamee, Paul  and
      P. McCrae, John  and
      Murray, Kenton  and
      Nadejde, Maria  and
      Nakamura, Satoshi  and
      Negri, Matteo  and
      Nguyen, Ha  and
      Niehues, Jan  and
      Niu, Xing  and
      Kr. Ojha, Atul  and
      E. Ortega, John  and
      Pal, Proyag  and
      Pino, Juan  and
      van der Plas, Lonneke  and
      Pol{\'a}k, Peter  and
      Rippeth, Elijah  and
      Salesky, Elizabeth  and
      Shi, Jiatong  and
      Sperber, Matthias  and
      St{\"u}ker, Sebastian  and
      Sudoh, Katsuhito  and
      Tang, Yun  and
      Thompson, Brian  and
      Tran, Kevin  and
      Turchi, Marco  and
      Waibel, Alex  and
      Wang, Mingxuan  and
      Watanabe, Shinji  and
      Zevallos, Rodolfo},
    editor = "Salesky, Elizabeth  and
      Federico, Marcello  and
      Carpuat, Marine",
    booktitle = "Proceedings of the 20th International Conference on Spoken Language Translation (IWSLT 2023)",
    month = jul,
    year = "2023",
    address = "Toronto, Canada (in-person and online)",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2023.iwslt-1.1",
    doi = "10.18653/v1/2023.iwslt-1.1",
    pages = "1--61",
    abstract = "This paper reports on the shared tasks organized by the 20th IWSLT Conference. The shared tasks address 9 scientific challenges in spoken language translation: simultaneous and offline translation, automatic subtitling and dubbing, speech-to-speech translation, multilingual, dialect and low-resource speech translation, and formality control. The shared tasks attracted a total of 38 submissions by 31 teams. The growing interest towards spoken language translation is also witnessed by the constantly increasing number of shared task organizers and contributors to the overview paper, almost evenly distributed across industry and academia.",
}