cvss.py

# Copyright 2022 The CVSS Dataset Authors and the dataset script contributor.
# All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# ==============================================================================
"""CVSS speech-to-speech translation corpus."""

import os
import datasets

_HOMEPAGE = 'https://github.com/google-research-datasets/cvss'

_DESCRIPTION = """\
CVSS is a massively multilingual-to-English speech-to-speech translation corpus,
covering sentence-level parallel speech-to-speech translation pairs from 21
languages into English.
"""

_CITATION = """\
@inproceedings{jia2022cvss,
    title={{CVSS} Corpus and Massively Multilingual Speech-to-Speech Translation},
    author={Jia, Ye and Tadmor Ramanovich, Michelle and Wang, Quan and Zen, Heiga},
    booktitle={Proceedings of Language Resources and Evaluation Conference (LREC)},
    pages={6691--6703},
    year={2022}
}
"""

_ROOT_URL = 'https://storage.googleapis.com/cvss'

_ALL_LANGUAGES = ('de', 'fr', 'es', 'ca', 'it', 'ru', 'zh', 'pt', 'fa', 'et',
                  'mn', 'nl', 'tr', 'ar', 'sv', 'lv', 'sl', 'ta', 'ja', 'id',
                  'cy')


def _get_download_urls(name='cvss_c', languages=_ALL_LANGUAGES, version='1.0'):
  """Gets URLs for downloading data.

  Args:
    name: 'cvss_c' or 'cvss_t'.
    languages: An iterable of source languages.
    version: Only '1.0' available.

  Returns:
    A list of URL strs.
  """
  return [
      f'{_ROOT_URL}/{name}_v{version}/{name}_{x}_en_v{version}.tar.gz'
      for x in languages
  ]


class CVSSConfig(datasets.BuilderConfig):
  """BuilderConfig for CVSS."""

  def __init__(self, name, languages='all', **kwargs):
    """BuilderConfig for CVSS.

    Args:
      name: 'cvss_c' or 'cvss_t'.
      languages: A list of source languages.
      **kwargs: keyword arguments forwarded to super.
    """
    super().__init__(name=name, **kwargs)
    if languages == 'all':
      self.languages = _ALL_LANGUAGES
    elif isinstance(languages, str):
      self.languages = [languages]
    else:
      self.languages = languages


class CVSS(datasets.GeneratorBasedBuilder):
  """CVSS dataset. Version 1.0."""

  BUILDER_CONFIG_CLASS = CVSSConfig

  VERSION = '1.0.0'

  DEFAULT_WRITER_BATCH_SIZE = 256

  def _info(self):
    return datasets.DatasetInfo(
        description=_DESCRIPTION,
        homepage=_HOMEPAGE,
        citation=_CITATION,
        features=datasets.Features({
            'id': datasets.Value('string'),
            'file': datasets.Value('string'),
            'audio': datasets.Audio(sampling_rate=24_000),
            'text': datasets.Value('string'),
        }))

  def _split_generators(self, dl_manager):
    print(self.config)
    downloaded_files = dl_manager.download_and_extract(
        _get_download_urls(self.config.name, self.config.languages))
    return [
        datasets.SplitGenerator(
            name=datasets.Split.TRAIN,
            gen_kwargs={
                'files': downloaded_files,
                'split': 'train'
            }),
        datasets.SplitGenerator(
            name=datasets.Split.VALIDATION,
            gen_kwargs={
                'files': downloaded_files,
                'split': 'dev'
            }),
        datasets.SplitGenerator(
            name=datasets.Split.TEST,
            gen_kwargs={
                'files': downloaded_files,
                'split': 'test'
            }),
    ]

  def _generate_examples(self, files, split):
    """Generates examples for each SplitGenerator."""
    for path in files:
      with open(os.path.join(path, f'{split}.tsv'), 'r', encoding='utf-8') as f:
        for line in f:
          cols = line.rstrip().split('\t')
          assert len(cols) == 2, cols
          key, text = cols

          audio_path = os.path.join(path, split, f'{key}.wav')
          with open(audio_path, 'rb') as audio_f:
            audio_bytes = audio_f.read()

          audio = {'path': audio_path, 'bytes': audio_bytes}
          yield key, {
              'id': key,
              'text': text,
              'audio': audio,
              'file': audio_path,
          }