Create ScreenTalk-XS.py

ScreenTalk-XS.py

@@ -0,0 +1,167 @@
+import os
+import json
+import csv
+import datasets
+from datasets.utils.py_utils import size_str
+from tqdm import tqdm
+
+from .languages import LANGUAGES
+from .release_stats import STATS
+
+_CITATION = """\
+@inproceedings{Trans,
+  title={fj11: A Massively Speech Corpus},
+  author={fj11},
+  year={2025}
+}
+"""
+
+_HOMEPAGE = "https://huggingface.co/datasets/DataLabX"
+
+_LICENSE = "CC-BY-SA-4.0"
+
+_BASE_URL = "https://huggingface.co/datasets/DataLabX/ScreenTalk-XS/resolve/main/"
+
+_AUDIO_URL = _BASE_URL + "audio/{lang}/{split}/{lang}_{split}_{shard_idx}.tar"
+
+_TRANSCRIPT_URL = _BASE_URL + "transcript/{lang}/{split}.tsv"
+
+_N_SHARDS_URL = _BASE_URL + "n_shards.json"
+
+
+def is_valid_token(token):
+    return True
+    
+
+class TransConfig(datasets.BuilderConfig):
+    """BuilderConfig for TransConfig."""
+    def __init__(self, name, version, **kwargs):
+        self.language = kwargs.pop("language", None)
+        self.release_date = kwargs.pop("release_date", None)
+        self.total_hr = kwargs.pop("total_hr", None)
+        self.validated_hr = kwargs.pop("validated_hr", None)
+        self.num_clips = kwargs.pop("num_clips", None)
+        self.size_bytes = kwargs.pop("size_bytes", None)
+        self.size_human = size_str(self.size_bytes)
+        description = (
+            f"This dataset consists of transcribed speech data from TV series and movies across various genres, "
+            f"including action, drama, sci-fi, and romance. It was released on {self.release_date} and contains "
+            f"{self.total_hr} hours of transcribed speech data. "
+            f"The dataset includes {self.num_clips} audio clips in {self.language}, with a total size of {self.size_human}, "
+            f"and is designed for automatic speech recognition (ASR) model training and fine-tuning, providing diverse and "
+            f"natural conversational speech from real-world entertainment media."
+        )
+
+        super(TransConfig, self).__init__(
+            name=name,
+            version=version,
+            description=description,
+            **kwargs
+        )
+
+
+class ScreenTalk(datasets.GeneratorBasedBuilder):
+    VERSION = datasets.Version("1.0.0")
+    BUILDER_CONFIGS = [
+        TransConfig(
+            name=lang,
+            version=STATS["version"],
+            language=LANGUAGES[lang],
+            release_date=STATS["date"],
+            num_clips=lang_stats["clips"],
+            total_hr=float(lang_stats["totalHrs"]) if lang_stats["totalHrs"] else None,
+            size_bytes=int(lang_stats["size"]) if lang_stats["size"] else None,
+        )
+        for lang, lang_stats in STATS["locales"].items()
+    ]
+
+    def _info(self):
+        total_languages = len(STATS["locales"])
+        total_valid_hours = STATS["totalValidHrs"]
+        description = (
+            f"This dataset consists of {total_valid_hours} hours of validated speech data "
+            f"in {total_languages} languages, with more content being continuously added. "
+            "It is designed for training and fine-tuning automatic speech recognition (ASR) models, "
+            "providing a diverse and realistic representation of spoken language."
+        )
+        features = datasets.Features(
+                {
+                    "audio": datasets.Audio(),  # Use datasets.Audio() instead of string
+                    "sentence": datasets.Value("string"),
+                }
+            )
+        return datasets.DatasetInfo(
+            description=description,
+            features=features,
+            supervised_keys=None,
+            homepage=_HOMEPAGE,
+            license=_LICENSE,
+            citation=_CITATION,
+            version=self.config.version,
+        )
+
+    def _split_generators(self, dl_manager):
+
+        lang = self.config.name
+        n_shards_path = dl_manager.download_and_extract(_N_SHARDS_URL)
+        with open(n_shards_path, encoding="utf-8") as f:
+            n_shards = json.load(f)
+
+        user_token = dl_manager.download_config.token
+        has_valid_token = is_valid_token(user_token)
+
+        audio_urls = {}
+        splits = ["xs"]
+
+        for split in splits:
+            audio_urls[split] = [
+                _AUDIO_URL.format(lang=lang, split=split, shard_idx=i) for i in range(n_shards[lang][split])
+            ]
+        archive_paths = dl_manager.download(audio_urls)
+        
+        local_extracted_archive_paths = dl_manager.extract(archive_paths) if not dl_manager.is_streaming else {}
+        meta_urls = {split: _TRANSCRIPT_URL.format(lang=lang, split=split) for split in splits}
+        meta_paths = dl_manager.download_and_extract(meta_urls)
+
+        split_generators = []
+        split_names = {
+            "xs": datasets.Split("xs"),
+        }
+
+        for split in splits:
+            split_generators.append(
+                datasets.SplitGenerator(
+                    name=split_names.get(split, split),
+                    gen_kwargs={
+                        "local_extracted_archive_paths": local_extracted_archive_paths.get(split),
+                        "archives": [dl_manager.iter_archive(path) for path in archive_paths.get(split)],
+                        "meta_path": meta_paths[split],
+                    },
+                ),
+            )
+
+        return split_generators
+
+    def _generate_examples(self, local_extracted_archive_paths, archives, meta_path):
+
+        metadata = {}
+        with open(meta_path, encoding="utf-8") as f:
+            reader = csv.DictReader(f, delimiter="\t", quoting=csv.QUOTE_NONE)
+            for row in tqdm(reader, desc="Reading metadata..."):
+                if not isinstance(row, dict):
+                    continue
+                path = row.get("audio")
+                _, filename = os.path.split(path)
+                metadata[filename] = 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
+                    # print("path: ", path)
+                    result["audio"] = path
+                    # result["path"] = path
+                    yield path, result