biulders.py

# Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
#
# 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.
"""Leading and Trailing Silences Removed Large Nepali ASR Dataset"""

import os
import csv

import datasets


_CITATION = """\
@inproceedings{kjartansson-etal-sltu2018,
title = {{Crowd-Sourced Speech Corpora for Javanese, Sundanese,  Sinhala, Nepali, and Bangladeshi Bengali}},
author = {Oddur Kjartansson and Supheakmungkol Sarin and Knot Pipatsrisawat and Martin Jansche and Linne Ha},
booktitle = {Proc. The 6th Intl. Workshop on Spoken Language Technologies for Under-Resourced Languages (SLTU)},
year  = {2018},
address = {Gurugram, India},
month = aug,
pages = {52--55},
URL   = {http://dx.doi.org/10.21437/SLTU.2018-11}
}
"""

_DESCRIPTION = """\
This data set contains transcribed audio data for Nepali. The data set consists of flac files, and a TSV file. The file utt_spk_text.tsv contains a FileID, anonymized UserID and the transcription of audio in the file.
The data set has been manually quality checked, but there might still be errors.
The audio files are sampled at rate of 16KHz, and leading and trailing silences are trimmed using torchaudio's voice activity detection.
"""

# Official homepage for the dataset
_HOMEPAGE = "https://www.openslr.org/54/"

# The licence for the dataset
_LICENSE = "license:cc-by-sa-4.0"

# 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/SumitMdhr/SANT-ASR/resolve/main/"
_URLS = {
    "zipfile": _URL + "CLEAN_DATA.zip",
    "index_file": _URL + "metedata1.csv",
}


# TODO: Name of the dataset usually match the script name with CamelCase instead of snake_case
class OpenslrNepaliAsrCleaned(datasets.GeneratorBasedBuilder):
    """End Silences Removed Large Nepali ASR Dataset"""

    VERSION = datasets.Version("1.0.0")
    # It's not mandatory to have a default configuration. Just use one if it make sense.
    DEFAULT_CONFIG_NAME = "original"

    def _info(self):
        features = datasets.Features(
            {
                "utterance_id": datasets.Value("string"),
                "speaker_id": datasets.Value("string"),
                "utterance": datasets.Audio(),
                "transcription": datasets.Value("string"),
                "num_frames": datasets.Value("int32"),
            }
        )
        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            # Here we define them above because they are different between the two configurations
            features=features,
            # If there's a common (input, target) tuple from the features, uncomment supervised_keys line below and
            # specify them. They'll be used if as_supervised=True in builder.as_dataset.
            # supervised_keys=("sentence", "label"),
            # Homepage of the dataset for documentation
            homepage=_HOMEPAGE,
            # License for the dataset if available
            license=_LICENSE,
            # Citation for the dataset
            citation=_CITATION,
            task_templates=[
                datasets.tasks.AutomaticSpeechRecognition(
                    audio_column="utterance", transcription_column="transcription"
                )
            ],
        )

    def _split_generators(self, dl_manager):
        index_file = dl_manager.download(_URLS["index_file"])
        zip_paths = dl_manager.download(_URLS["zipfiles"])
        audio_paths = dl_manager.extract(zip_paths)
        for path in zip_paths:
            if os.path.exists(path):
                os.remove(path)
        return [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                gen_kwargs={
                    "index_file": index_file,
                    "audio_paths": audio_paths,
                },
            ),
        ]

    def _generate_examples(self, index_file, audio_paths):
        with open(index_file, encoding="utf-8") as f:
            reader = csv.DictReader(f, delimiter="\t")
            for key, row in enumerate(reader):
                path = os.path.join(audio_paths, "CLEAN_DATA", row["utterance_id"])
                yield key, {
                    "utterance_id": row["utterance_id"],
                    "speaker_id": row["speaker_id"],
                    "utterance": path,
                    "transcription": row["transcription"],
                    "num_frames": int(row["num_frames"]),
                }