Uyghur_asr.py

import datasets
from datasets import DownloadManager, DatasetInfo
from datasets.tasks import AutomaticSpeechRecognition
import os
import csv
import json

# TODO: Add BibTeX citation
# Find for instance the citation on arxiv or on the dataset repo/website
_CITATION = """\
@InProceedings{huggingface:dataset,
title = {A great new dataset},
author={huggingface, Inc.
},
year={2022}
}
"""
# TODO: Add description of the dataset here
# You can copy an official description
_DESCRIPTION = """\
This new dataset is from Xinjiang University and to do some ASR in low resource.
"""

# TODO: Add a link to an official homepage for the dataset here
_HOMEPAGE = "http://www.openslr.org/22"

# TODO: Add the licence for the dataset here if you can find it
_LICENSE = ""

# TODO: Add link to the official dataset URLs here
# The HuggingFace Datasets library doesn't host the datasets but only points to the original files.
# This can be an arbitrary nested dict/list of URLs (see below in `_split_generators` method)

_URL = "https://huggingface.co/datasets/sunlixu/Uyghur/"

_DL_URLS = {
    "train": 'https://cdn-lfs.huggingface.co/repos/69/21/6921bf0087a98e7fb471f5c8b64282577d7fcb856d37084e0b482955ffe6e1ae/f06c4e7e4d8f36efda4cd5e487a4a7e573965aa987f6dc138684dd7f13c12ed1?response-content-disposition=attachment%3B%20filename%3D%22train.zip%22',
    "test": "https://cdn-lfs.huggingface.co/repos/69/21/6921bf0087a98e7fb471f5c8b64282577d7fcb856d37084e0b482955ffe6e1ae/92b3ffbc5c3b4d40d737f8b2995f27bcbce2107b44d06aada08df9f7ba2c58ed?response-content-disposition=attachment%3B%20filename%3D%22test.zip%22",
    "cv": "https://cdn-lfs.huggingface.co/repos/69/21/6921bf0087a98e7fb471f5c8b64282577d7fcb856d37084e0b482955ffe6e1ae/2d034279c7453393a61abd4163ecdd0d06c49ede1f71f0a347e3212c8de38da8?response-content-disposition=attachment%3B%20filename%3D%22cv.zip%22",
    "all": {
        "train": "https://cdn-lfs.huggingface.co/repos/69/21/6921bf0087a98e7fb471f5c8b64282577d7fcb856d37084e0b482955ffe6e1ae/f06c4e7e4d8f36efda4cd5e487a4a7e573965aa987f6dc138684dd7f13c12ed1?response-content-disposition=attachment%3B%20filename%3D%22train.zip%22",
        "test": "https://cdn-lfs.huggingface.co/repos/69/21/6921bf0087a98e7fb471f5c8b64282577d7fcb856d37084e0b482955ffe6e1ae/92b3ffbc5c3b4d40d737f8b2995f27bcbce2107b44d06aada08df9f7ba2c58ed?response-content-disposition=attachment%3B%20filename%3D%22test.zip%22",
        "cv": "https://cdn-lfs.huggingface.co/repos/69/21/6921bf0087a98e7fb471f5c8b64282577d7fcb856d37084e0b482955ffe6e1ae/2d034279c7453393a61abd4163ecdd0d06c49ede1f71f0a347e3212c8de38da8?response-content-disposition=attachment%3B%20filename%3D%22cv.zip%22"
    },
}


class UyghurASRConfig(datasets.BuilderConfig):
    def __init__(self, **kwargs):
        """BuilderConfig for SQUAD.
        Args:
          **kwargs: keyword arguments forwarded to super.
        """
        super(UyghurASRConfig, self).__init__(**kwargs)


class UyghurASR(datasets.GeneratorBasedBuilder):
    DEFAULT_WRITER_BATCH_SIZE = 256
    DEFAULT_CONFIG_NAME = "all"
    BUILDER_CONFIGS = [
        UyghurASRConfig(name="train", description="'train' speech."),
        UyghurASRConfig(name="test", description="'test' speech."),
        UyghurASRConfig(name="cv", description="'cv' speech."),
        UyghurASRConfig(name="all", description="all"),
    ]

    def _info(self) -> DatasetInfo:
        return datasets.DatasetInfo(
            description="Uyghur_20",
            features=datasets.Features(
                {
                    "file": datasets.Value("string"),
                    "audio": datasets.features.Audio(sampling_rate=16_000),
                    "text": datasets.Value("string"),
                    "speaker_id": datasets.Value("int64"),
                    "id": datasets.Value("string"),
                }
            ),
            supervised_keys=None,
            homepage='https://huggingface.co/datasets/sunlixu/Uyghur/',
            citation=_CITATION,
            task_templates=[AutomaticSpeechRecognition(audio_column="audio", transcription_column="text")],
        )

    def _split_generators(self, dl_manager: DownloadManager):
        archive_path = dl_manager.download(_DL_URLS[self.config.name])
        local_extracted_archive = dl_manager.extract(archive_path) if not dl_manager.is_streaming else {}

        return [
            datasets.SplitGenerator(name=datasets.Split.TRAIN,
                                    gen_kwargs={"local_extracted_archive": local_extracted_archive.get('train'),
                                                "files": dl_manager.iter_archive(archive_path["train"]),
                                                },
                                    ),
            datasets.SplitGenerator(name=datasets.Split.TEST,
                                    gen_kwargs={"local_extracted_archive": local_extracted_archive.get('test'),
                                                "files": dl_manager.iter_archive(archive_path["test"]),
                                                },
                                    ),
            datasets.SplitGenerator(name=datasets.Split.VALIDATION,
                                    gen_kwargs={"local_extracted_archive": local_extracted_archive.get('cv'),
                                                "files": dl_manager.iter_archive(archive_path["cv"]),
                                                },
                                    )
        ]

    def _generate_examples(self, files, local_extracted_archive):
        """Generate examples from a Uyghur archive_path."""
        key = 0
        audio_data = {}
        transcripts = []
        for path, f in files:
            if path.endswith(".wav"):
                id_ = path.split("/")[-1][: -len(".wav")]
                audio_data[id_] = f.read()
            elif path.endswith(".txt"):
                for line in f:
                    if line:
                        line = line.decode("utf-8").strip()
                        id_, transcript = line.split(" ", 1)
                        audio_file = f"{id_}.wav"
                        # speaker_id, chapter_id = [int(el) for el in id_.split("-")[:2]]
                        speaker_id = id_.split('_')[0]
                        audio_file = (
                            os.path.join(local_extracted_archive, audio_file)
                            if local_extracted_archive
                            else audio_file
                        )
                        transcripts.append(
                            {
                                "id": id_,
                                "speaker_id": speaker_id,
                                # "chapter_id": chapter_id,
                                "file": audio_file,
                                "text": transcript,
                            }
                        )
            if audio_data and len(audio_data) == len(transcripts):
                for transcript in transcripts:
                    audio = {"path": transcript["file"], "bytes": audio_data[transcript["id"]]}
                    yield key, {"audio": audio, **transcript}
                    key += 1
                audio_data = {}
                transcripts = []