librispeech_asr

Referências:

limpar

Use o seguinte comando para carregar este conjunto de dados no TFDS:

ds = tfds.load('huggingface:librispeech_asr/clean')
  • Descrição :
LibriSpeech is a corpus of approximately 1000 hours of read English speech with sampling rate of 16 kHz,
prepared by Vassil Panayotov with the assistance of Daniel Povey. The data is derived from read
audiobooks from the LibriVox project, and has been carefully segmented and aligned.87

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .flac 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 soundfile as sf

def map_to_array(batch):
    speech_array, _ = sf.read(batch["file"])
    batch["speech"] = speech_array
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licença : Nenhuma licença conhecida
  • Versão : 2.1.0
  • Divisões :
Dividir Exemplos
'test' 2620
'train.100' 28539
'train.360' 104014
'validation' 2703
  • Características :
{
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "audio": {
        "sampling_rate": 16000,
        "mono": true,
        "decode": 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"
    }
}

outro

Use o seguinte comando para carregar este conjunto de dados no TFDS:

ds = tfds.load('huggingface:librispeech_asr/other')
  • Descrição :
LibriSpeech is a corpus of approximately 1000 hours of read English speech with sampling rate of 16 kHz,
prepared by Vassil Panayotov with the assistance of Daniel Povey. The data is derived from read
audiobooks from the LibriVox project, and has been carefully segmented and aligned.87

Note that in order to limit the required storage for preparing this dataset, the audio
is stored in the .flac 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 soundfile as sf

def map_to_array(batch):
    speech_array, _ = sf.read(batch["file"])
    batch["speech"] = speech_array
    return batch

dataset = dataset.map(map_to_array, remove_columns=["file"])
  • Licença : Nenhuma licença conhecida
  • Versão : 2.1.0
  • Divisões :
Dividir Exemplos
'test' 2939
'train.500' 148688
'validation' 2864
  • Características :
{
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "audio": {
        "sampling_rate": 16000,
        "mono": true,
        "decode": 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"
    }
}

todos

Use o seguinte comando para carregar este conjunto de dados no TFDS:

ds = tfds.load('huggingface:librispeech_asr/all')
  • Descrição :
LibriSpeech is a corpus of approximately 1000 hours of read English speech with sampling rate of 16 kHz,
prepared by Vassil Panayotov with the assistance of Daniel Povey. The data is derived from read
audiobooks from the LibriVox project, and has been carefully segmented and aligned.87
  • Licença : Nenhuma licença conhecida
  • Versão : 2.1.0
  • Divisões :
Dividir Exemplos
'test.clean' 2620
'test.other' 2939
'train.clean.100' 28539
'train.clean.360' 104014
'train.other.500' 148688
'validation.clean' 2703
'validation.other' 2864
  • Características :
{
    "file": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "audio": {
        "sampling_rate": 16000,
        "mono": true,
        "decode": 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"
    }
}