CORAA-MUPE-ASR.py

import csv
import datasets

_PROMPTS_URLS = {
    "train": "train.csv",
    "validation": "validation.csv",
    "test": "test.csv",
}

_ARCHIVES = {
    "train": "train.tar.gz",
    "validation": "validation.tar.gz",
    "test": "test.tar.gz",
}   

_PATH_TO_CLIPS = {
    "train": "train",
    "validation": "validation",
    "test": "test",
}

class MUPEASRDataset(datasets.GeneratorBasedBuilder):
    def _info(self):
        return datasets.DatasetInfo(
            features=datasets.Features(
                {
                    "audio_id": datasets.Value("int64"),
                    "audio_name": datasets.Value("string"),
                    "file_path": datasets.Value("string"),
                    "speaker_type": datasets.Value("string"),
                    "speaker_code": datasets.Value("string"),
                    "speaker_gender": datasets.Value("string"),
                    "education": datasets.Value("string"),
                    "birth_state": datasets.Value("string"),
                    "birth_country": datasets.Value("string"),
                    "age": datasets.Value("int64"),
                    "birth_year": datasets.Value("int64"),
                    "audio_quality": datasets.Value("string"),
                    "start_time": datasets.Value("float32"),
                    "end_time": datasets.Value("float32"),
                    "duration": datasets.Value("float32"),
                    "normalized_text": datasets.Value("string"),
                    "original_text": datasets.Value("string"),
                    "audio": datasets.Audio(sampling_rate=16_000),
                }
            )
        )

    def _split_generators(self, dl_manager):
        prompts_path = dl_manager.download(_PROMPTS_URLS)
        archive = dl_manager.download(_ARCHIVES)

        return [
            datasets.SplitGenerator(
                name = datasets.Split.VALIDATION,
                gen_kwargs = {
                    "prompts_path": prompts_path["validation"],
                    "path_to_clips": _PATH_TO_CLIPS["validation"],
                    "audio_files": dl_manager.iter_archive(archive["validation"]),
                }
            ),
            datasets.SplitGenerator(
                name = datasets.Split.TEST,
                gen_kwargs = {
                    "prompts_path": prompts_path["test"],
                    "path_to_clips": _PATH_TO_CLIPS["test"],
                    "audio_files": dl_manager.iter_archive(archive["test"]),
                }
            ),
            datasets.SplitGenerator(
                name = datasets.Split.TRAIN,
                gen_kwargs = {
                    "prompts_path": prompts_path["train"],
                    "path_to_clips": _PATH_TO_CLIPS["train"],
                    "audio_files": dl_manager.iter_archive(archive["train"]),
                }
            ),
        ]

    def _generate_examples(self, prompts_path, path_to_clips, audio_files):
        examples = {}
        with open(prompts_path, "r") as f:
            csv_reader = csv.DictReader(f)
            for row in csv_reader:
                audio_id = row['audio_id']
                audio_name = row['audio_name']
                file_path = row['file_path']
                speaker_type = row['speaker_type']
                speaker_code = row['speaker_code']
                speaker_gender = row['speaker_gender']
                education = row['education']
                birth_state = row['birth_state']
                birth_country = row['birth_country']
                age = row['age']
                birth_year = row['birth_year']
                audio_quality = row['audio_quality']
                start_time = row['start_time']
                end_time = row['end_time']
                duration = row['duration']
                normalized_text = row['normalized_text']
                original_text = row['original_text']

                examples[file_path] = {
                    "audio_id" : audio_id,
                    "audio_name" : audio_name,
                    "file_path" : file_path,
                    "speaker_type" : speaker_type,
                    "speaker_code" : speaker_code,
                    "speaker_gender" : speaker_gender,
                    "education" : education,
                    "birth_state" : birth_state,
                    "birth_country" : birth_country,
                    "age" : age,
                    "birth_year" : birth_year,
                    "audio_quality" : audio_quality,
                    "start_time" : start_time,
                    "end_time" : end_time,
                    "duration" : duration,
                    "normalized_text" : normalized_text,
                    "original_text" : original_text,
                }
        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