тедлиум

  • Описание :

Корпус TED-LIUM представляет собой выступления TED на английском языке с транскрипцией, сэмплированные на частоте 16 кГц. Он содержит около 118 часов речи.

FeaturesDict({
   
'gender': ClassLabel(shape=(), dtype=int64, num_classes=3),
   
'id': string,
   
'speaker_id': string,
   
'speech': Audio(shape=(None,), dtype=int64),
   
'text': Text(shape=(), dtype=string),
})
  • Документация по функциям :
Особенность Сорт Форма Dтип Описание
ОсобенностиDict
пол Метка класса int64
идентификатор Тензор нить
спикер_ид Тензор нить
речь Аудио (Никто,) int64
текст Текст нить
  • Контролируемые ключи (см. as_supervised doc ): ('speech', 'text')

  • Рисунок ( tfds.show_examples ): не поддерживается.

  • Цитата :

@inproceedings{rousseau2012tedlium,
  title
={TED-LIUM: an Automatic Speech Recognition dedicated corpus},
  author
={Rousseau, Anthony and Del{\'e}glise, Paul and Est{\`e}ve, Yannick},
  booktitle={Conference on Language Resources and Evaluation (LREC)},
  pages={125--129},
  year={2012}
}

tedlium/release1 (конфигурация по умолчанию)

  • Описание конфигурации : Корпус TED-LIUM представляет собой англоязычные выступления TED с транскрипцией, сэмплированные на частоте 16 кГц. Он содержит около 118 часов речи.

    This is the TED-LIUM corpus release 1,
    licensed under
    Creative Commons BY-NC-ND 3.0
    (http://creativecommons.org/licenses/by-nc-nd/3.0/deed.en).
  • Домашняя страница : https://www.openslr.org/7/

  • Размер загрузки : 19.82 GiB

  • Размер набора данных : 39.23 GiB

  • Сплиты :

Расколоть Примеры
'test' 1469
'train' 56 803
'validation' 591

Тедлиум/выпуск2

  • Описание конфигурации : это выпуск 2 корпуса TED-LIUM, лицензированный Creative Commons BY-NC-ND 3.0 ( http://creativecommons.org/licenses/by-nc-nd/3.0/deed.en ).

    All talks and text are property of TED Conferences LLC.

    The TED-LIUM corpus was made from audio talks and their transcriptions
    available on the TED website
    . We have prepared and filtered these data
    in order to train acoustic models to participate to the International
    Workshop on Spoken Language Translation 2011 (the LIUM English/French
    SLT system reached the first rank
    in the SLT task).

    Contains 1495 talks and transcripts.
  • Домашняя страница : https://www.openslr.org/19/

  • Размер загрузки : 34.26 GiB

  • Размер набора данных : 67.04 GiB

  • Сплиты :

Расколоть Примеры
'test' 1469
'train' 92 973
'validation' 591

Тедлиум/релиз3

  • Описание конфигурации : это выпуск 3 корпуса TED-LIUM, лицензированный Creative Commons BY-NC-ND 3.0.

    All talks and text are property of TED Conferences LLC.

    This new TED-LIUM release was made through a collaboration between the
    Ubiqus company and the LIUM (University of Le Mans, France)

    Contents:

    - 2351 audio talks in NIST sphere format (SPH), including talks from
      TED
    -LIUM 2: be careful, same talks but not same audio files (only
      these audio file must be used
    with the TED-LIUM 3 STM files)
    - 452 hours of audio
    - 2351 aligned automatic transcripts in STM format
    - TEDLIUM 2 dev and test data: 19 TED talks in SPH format with
      corresponding manual transcriptions
    (cf. 'legacy' distribution below).
    - Dictionary with pronunciations (159848 entries), same file as the one
      included
    in TED-LIUM 2
    - Selected monolingual data for language modeling from WMT12 publicly
      available corpora
    : these files come from the TED-LIUM 2 release, but
      have been modified to
    get a tokenization more relevant for English
      language

    Two corpus distributions:
    - the legacy one, on which the dev and test datasets are the same as in
      TED
    -LIUM 2 (and TED-LIUM 1).
    - the 'speaker adaptation' one, especially designed for experiments on
      speaker adaptation
    .
  • Домашняя страница : https://www.openslr.org/51/

  • Размер загрузки : 50.59 GiB

  • Размер набора данных : 145.67 GiB

  • Сплиты :

Расколоть Примеры
'test' 1469
'train' 268 263
'validation' 591