covost2

Les références:

fr_de

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/en_de')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 15531
'train' 289430
'validation' 15531
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

fr_tr

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/en_tr')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 15531
'train' 289430
'validation' 15531
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

fr_fa

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/en_fa')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 15531
'train' 289430
'validation' 15531
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

fr_sv-SE

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/en_sv-SE')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 15531
'train' 289430
'validation' 15531
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

fr_mn

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/en_mn')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 15531
'train' 289430
'validation' 15531
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

fr_zh-CN

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/en_zh-CN')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 15531
'train' 289430
'validation' 15531
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

fr_cy

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/en_cy')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 15531
'train' 289430
'validation' 15531
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

fr_ca

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/en_ca')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 15531
'train' 289430
'validation' 15531
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

fr_sl

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/en_sl')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 15531
'train' 289430
'validation' 15531
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

fr_et

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/en_et')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 15531
'train' 289430
'validation' 15531
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

fr_id

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/en_id')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 15531
'train' 289430
'validation' 15531
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

fr_ar

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/en_ar')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 15531
'train' 289430
'validation' 15531
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

fr_ta

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/en_ta')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 15531
'train' 289430
'validation' 15531
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

fr_lv

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/en_lv')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 15531
'train' 289430
'validation' 15531
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

fr_ja

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/en_ja')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 15531
'train' 289430
'validation' 15531
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

fr_en

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/fr_en')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 14760
'train' 207374
'validation' 14760
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

de_fr

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/de_en')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 13511
'train' 127834
'validation' 13511
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

es_fr

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/es_en')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 13221
'train' 79015
'validation' 13221
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

ca_fr

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/ca_en')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 12730
'train' 95854
'validation' 12730
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

it_fr

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/it_en')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 8951
'train' 31698
'validation' 8940
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

ru_fr

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/ru_en')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 6300
'train' 12112
'validation' 6110
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

zh-CN_fr

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/zh-CN_en')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 4898
'train' 7085
'validation' 4843
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

pt_fr

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/pt_en')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 4023
'train' 9158
'validation' 3318
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

fa_fr

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/fa_en')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 3445
'train' 53949
'validation' 3445
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

et_fr

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/et_en')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 1571
'train' 1782
'validation' 1576
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

mn_fr

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/mn_en')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 1759
'train' 2067
'validation' 1761
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

nl_fr

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/nl_en')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 1699
'train' 7108
'validation' 1699
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

tr_en

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/tr_en')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 1629
'train' 3966
'validation' 1624
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

ar_en

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/ar_en')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 1695
'train' 2283
'validation' 1758
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

sv-SE_fr

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/sv-SE_en')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 1595
'train' 2160
'validation' 1349
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

lv_fr

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/lv_en')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 1629
'train' 2337
'validation' 1125
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

sl_fr

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/sl_en')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 360
'train' 1843
'validation' 509
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

ta_fr

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/ta_en')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 786
'train' 1358
'validation' 384
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

ja_en

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/ja_en')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 684
'train' 1119
'validation' 635
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

id_fr

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/id_en')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 844
'train' 1243
'validation' 792
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

cy_fr

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:covost2/cy_en')
  • Description :
CoVoST 2, a large-scale multilingual speech translation corpus covering translations from 21 languages into English and from English into 15 languages. The dataset is created using Mozilla’s open source Common Voice database of crowdsourced voice recordings.

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .mp3 format and is not converted to a float32 array. To convert, the audio
file to a float32 array, please make use of the `.map()` function as follows:


python
import torchaudio

def map_to_array(batch):
    speech_array, _ = torchaudio.load(batch["file"])
    batch["speech"] = speech_array.numpy()
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licence : Aucune licence connue
  • Version : 1.0.0
  • Divisions :
Diviser Exemples
'test' 690
'train' 1241
'validation' 690
  • Caractéristiques :
{
    "client_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "translation": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}