vivos_noisy.py

import datasets

_DATA_URL = "data/vivos_noisy.tar.gz"

_PROMPTS_URLS = {
    "train": "data/train_prompts.txt.gz",
    "test": "data/test_prompts.txt.gz",
}


class VivosNoisyDataset(datasets.GeneratorBasedBuilder):
    """VIVOS NOISY is a Vietnamese speech corpus with added noise, based on the original VIVOS dataset.
    This corpus is prepared for Vietnamese Automatic Speech Recognition task under noisy environments."""

    VERSION = datasets.Version("1.1.0")

    def _info(self):
        return datasets.DatasetInfo(
            # This is the description that will appear on the datasets page.
            description=_DESCRIPTION,
            features=datasets.Features(
                {
                    "speaker_id": datasets.Value("string"),
                    "path": datasets.Value("string"),
                    "audio": datasets.Audio(sampling_rate=16_000),
                    "sentence": datasets.Value("string"),
                }
            ),
            supervised_keys=None,
        )

    def _split_generators(self, dl_manager):
        prompts_paths = dl_manager.download_and_extract(_PROMPTS_URLS)
        archive = dl_manager.download(_DATA_URL)
        train_dir = "vivos_noisy/train"
        test_dir = "vivos_noisy/test"

        return [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                # These kwargs will be passed to _generate_examples
                gen_kwargs={
                    "prompts_path": prompts_paths["train"],
                    "path_to_clips": train_dir + "/waves",
                    "audio_files": dl_manager.iter_archive(archive),
                },
            ),
            datasets.SplitGenerator(
                name=datasets.Split.TEST,
                # These kwargs will be passed to _generate_examples
                gen_kwargs={
                    "prompts_path": prompts_paths["test"],
                    "path_to_clips": test_dir + "/waves",
                    "audio_files": dl_manager.iter_archive(archive),
                },
            ),
        ]

    def _generate_examples(self, prompts_path, path_to_clips, audio_files):
        """Yields examples as (key, example) tuples."""
        # This method handles input defined in _split_generators to yield (key, example) tuples from the dataset.
        # The `key` is here for legacy reason (tfds) and is not important in itself.
        examples = {}
        with open(prompts_path, encoding="utf-8") as f:
            for row in f:
                data = row.strip().split(" ", 1)
                # Extract speaker_id from the full filename
                # For example: VIVOS_NOISY_VIVOSDEV01_R002_001 -> VIVOS_NOISY_VIVOSDEV01
                filename_parts = data[0].split("_")
                if len(filename_parts) >= 3:
                    # Join the first 3 parts to get the speaker_id (VIVOS_NOISY_VIVOSDEV01)
                    speaker_id = "_".join(filename_parts[:3])
                else:
                    # Fallback if the naming convention is different
                    speaker_id = filename_parts[0]
                
                audio_path = "/".join([path_to_clips, speaker_id, data[0] + ".wav"])
                examples[audio_path] = {
                    "speaker_id": speaker_id,
                    "path": audio_path,
                    "sentence": data[1],
                }
        
        inside_clips_dir = False
        id_ = 0
        for path, f in audio_files:
            if path.startswith(path_to_clips):
                inside_clips_dir = True
                if path in examples:
                    audio = {"path": path, "bytes": f.read()}
                    yield id_, {**examples[path], "audio": audio}
                    id_ += 1
            elif inside_clips_dir:
                break