multilingual_librispeech

References:

polish

Use the following command to load this dataset in TFDS:

ds = tfds.load('huggingface:multilingual_librispeech/polish')
  • Description:
Multilingual LibriSpeech (MLS) dataset is a large multilingual corpus suitable for speech research. The dataset is derived from read audiobooks from LibriVox and consists of 8 languages - English, German, Dutch, Spanish, French, Italian, Portuguese, Polish.
  • License: No known license
  • Version: 2.1.0
  • Splits:
Split Examples
'test' 520
'train' 25043
'train.1h' 238
'train.9h' 2173
'validation' 512
  • Features:
{
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "audio": {
        "sampling_rate": 16000,
        "mono": true,
        "id": null,
        "_type": "Audio"
    },
    "text": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "speaker_id": {
        "dtype": "int64",
        "id": null,
        "_type": "Value"
    },
    "chapter_id": {
        "dtype": "int64",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

german

Use the following command to load this dataset in TFDS:

ds = tfds.load('huggingface:multilingual_librispeech/german')
  • Description:
Multilingual LibriSpeech (MLS) dataset is a large multilingual corpus suitable for speech research. The dataset is derived from read audiobooks from LibriVox and consists of 8 languages - English, German, Dutch, Spanish, French, Italian, Portuguese, Polish.
  • License: No known license
  • Version: 2.1.0
  • Splits:
Split Examples
'test' 3394
'train' 469942
'train.1h' 241
'train.9h' 2194
'validation' 3469
  • Features:
{
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "audio": {
        "sampling_rate": 16000,
        "mono": true,
        "id": null,
        "_type": "Audio"
    },
    "text": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "speaker_id": {
        "dtype": "int64",
        "id": null,
        "_type": "Value"
    },
    "chapter_id": {
        "dtype": "int64",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

dutch

Use the following command to load this dataset in TFDS:

ds = tfds.load('huggingface:multilingual_librispeech/dutch')
  • Description:
Multilingual LibriSpeech (MLS) dataset is a large multilingual corpus suitable for speech research. The dataset is derived from read audiobooks from LibriVox and consists of 8 languages - English, German, Dutch, Spanish, French, Italian, Portuguese, Polish.
  • License: No known license
  • Version: 2.1.0
  • Splits:
Split Examples
'test' 3075
'train' 374287
'train.1h' 234
'train.9h' 2153
'validation' 3095
  • Features:
{
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "audio": {
        "sampling_rate": 16000,
        "mono": true,
        "id": null,
        "_type": "Audio"
    },
    "text": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "speaker_id": {
        "dtype": "int64",
        "id": null,
        "_type": "Value"
    },
    "chapter_id": {
        "dtype": "int64",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

french

Use the following command to load this dataset in TFDS:

ds = tfds.load('huggingface:multilingual_librispeech/french')
  • Description:
Multilingual LibriSpeech (MLS) dataset is a large multilingual corpus suitable for speech research. The dataset is derived from read audiobooks from LibriVox and consists of 8 languages - English, German, Dutch, Spanish, French, Italian, Portuguese, Polish.
  • License: No known license
  • Version: 2.1.0
  • Splits:
Split Examples
'test' 2426
'train' 258213
'train.1h' 241
'train.9h' 2167
'validation' 2416
  • Features:
{
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "audio": {
        "sampling_rate": 16000,
        "mono": true,
        "id": null,
        "_type": "Audio"
    },
    "text": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "speaker_id": {
        "dtype": "int64",
        "id": null,
        "_type": "Value"
    },
    "chapter_id": {
        "dtype": "int64",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

spanish

Use the following command to load this dataset in TFDS:

ds = tfds.load('huggingface:multilingual_librispeech/spanish')
  • Description:
Multilingual LibriSpeech (MLS) dataset is a large multilingual corpus suitable for speech research. The dataset is derived from read audiobooks from LibriVox and consists of 8 languages - English, German, Dutch, Spanish, French, Italian, Portuguese, Polish.
  • License: No known license
  • Version: 2.1.0
  • Splits:
Split Examples
'test' 2385
'train' 220701
'train.1h' 233
'train.9h' 2110
'validation' 2408
  • Features:
{
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "audio": {
        "sampling_rate": 16000,
        "mono": true,
        "id": null,
        "_type": "Audio"
    },
    "text": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "speaker_id": {
        "dtype": "int64",
        "id": null,
        "_type": "Value"
    },
    "chapter_id": {
        "dtype": "int64",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

italian

Use the following command to load this dataset in TFDS:

ds = tfds.load('huggingface:multilingual_librispeech/italian')
  • Description:
Multilingual LibriSpeech (MLS) dataset is a large multilingual corpus suitable for speech research. The dataset is derived from read audiobooks from LibriVox and consists of 8 languages - English, German, Dutch, Spanish, French, Italian, Portuguese, Polish.
  • License: No known license
  • Version: 2.1.0
  • Splits:
Split Examples
'test' 1262
'train' 59623
'train.1h' 240
'train.9h' 2173
'validation' 1248
  • Features:
{
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "audio": {
        "sampling_rate": 16000,
        "mono": true,
        "id": null,
        "_type": "Audio"
    },
    "text": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "speaker_id": {
        "dtype": "int64",
        "id": null,
        "_type": "Value"
    },
    "chapter_id": {
        "dtype": "int64",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

portuguese

Use the following command to load this dataset in TFDS:

ds = tfds.load('huggingface:multilingual_librispeech/portuguese')
  • Description:
Multilingual LibriSpeech (MLS) dataset is a large multilingual corpus suitable for speech research. The dataset is derived from read audiobooks from LibriVox and consists of 8 languages - English, German, Dutch, Spanish, French, Italian, Portuguese, Polish.
  • License: No known license
  • Version: 2.1.0
  • Splits:
Split Examples
'test' 871
'train' 37533
'train.1h' 236
'train.9h' 2116
'validation' 826
  • Features:
{
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "audio": {
        "sampling_rate": 16000,
        "mono": true,
        "id": null,
        "_type": "Audio"
    },
    "text": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "speaker_id": {
        "dtype": "int64",
        "id": null,
        "_type": "Value"
    },
    "chapter_id": {
        "dtype": "int64",
        "id": null,
        "_type": "Value"
    },
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}