youtube-persian-asr.py

import os
import tarfile
import datasets
import pandas as pd
from typing import Dict, List
import io
from tqdm import tqdm
import csv
import os

_DESCRIPTION = """
This dataset consists of over 385 hours of audio extracted from various YouTube videos in the Persian language.

Note: This dataset contains raw, unvalidated transcriptions. Users are advised to:
1. Perform their own quality assessment
2. Create their own train/validation/test splits based on their specific needs
3. Validate a subset of the data if needed for their use case
"""

_CITATION = """
Use this repo info/link for citation.
"""

_LICENSE = "https://creativecommons.org/publicdomain/zero/1.0/"

_HOMEPAGE = "https://huggingface.co/datasets/PerSets/youtube-persian-asr"

_BASE_URL = "https://huggingface.co/datasets/PerSets/youtube-persian-asr/resolve/main/"

_AUDIO_URL = _BASE_URL + "data/unvalidated_{shard_idx}.tar"

class FarsiYoutubeDataset(datasets.GeneratorBasedBuilder):
    
    DEFAULT_WRITER_BATCH_SIZE = 1000

    VERSION = datasets.Version("1.0.0")

    def _info(self):
        return datasets.DatasetInfo(
            features=datasets.Features({
                "audio": datasets.Audio(sampling_rate=44_000),  # Adjust sampling rate as needed
                "text": datasets.Value("string"),
                "file_name": datasets.Value("string"),
            }),
            supervised_keys=None,
            license=_LICENSE,
            citation=_CITATION,
            version=self.VERSION,
            description=_DESCRIPTION
        )

    def _split_generators(self, dl_manager):
        """Returns SplitGenerators."""
        
        archive_paths = [_AUDIO_URL.format(shard_idx=i) for i in range(1, 22)]
        local_extracted_archive_paths = dl_manager.extract(archive_paths) if not dl_manager.is_streaming else {}

        return [
            datasets.SplitGenerator(
                name="unvalidated",  # Or adjust splits as needed
                gen_kwargs={
                    #"tar_dir": tar_dir,
                    #"metadata_path": metadata_path,
                    "local_extracted_archive_paths": local_extracted_archive_paths,
                    "archives": [dl_manager.iter_archive(path) for path in archive_paths],
                    "meta_path": _BASE_URL + "unvalidated.csv",
                },
            ),
        ]

    def _generate_examples(self, local_extracted_archive_paths, archives, meta_path):
        """Yields examples."""
        # Load TSV metadata
        data_fields = list(self._info().features.keys())
        metadata = {}
        with open(meta_path, encoding="utf-8") as f:
            reader = csv.DictReader(f, delimiter=",", quoting=csv.QUOTE_NONE)
            for row in tqdm(reader, desc="Reading metadata..."):
                if not row["file_name"].endswith(".mp3"):
                    row["file_name"] += ".mp3"
                if "sentence" in row:
                    row['text'] = row['sentence']
                    del row['sentence']
                for field in data_fields:
                    if field not in row:
                        row[field] = ""
                metadata[row["file_name"]] = row

        for i, audio_archive in enumerate(archives):
            for path, file in audio_archive:
                _, filename = os.path.split(path)
                if filename in metadata:
                    result = dict(metadata[filename])
                    # set the audio feature and the path to the extracted file
                    path = os.path.join(local_extracted_archive_paths[i], path) if local_extracted_archive_paths else path
                    result["audio"] = {"path": path, "bytes": file.read()}
                    result["file_name"] = path
                    yield path, result