bel_canto.py

import os
import random
import hashlib
import datasets
from datasets.tasks import ImageClassification


_NAMES = {
    "all": ["m_bel", "f_bel", "m_folk", "f_folk"],
    "gender": ["female", "male"],
    "singing_method": ["Folk_Singing", "Bel_Canto"],
}

_HOMEPAGE = f"https://www.modelscope.cn/datasets/ccmusic-database/{os.path.basename(__file__)[:-3]}"

_DOMAIN = f"{_HOMEPAGE}/resolve/master/data"

_URLS = {
    "audio": f"{_DOMAIN}/audio.zip",
    "mel": f"{_DOMAIN}/mel.zip",
    "eval": f"{_DOMAIN}/eval.zip",
}


class bel_canto(datasets.GeneratorBasedBuilder):
    def _info(self):
        return datasets.DatasetInfo(
            features=(
                datasets.Features(
                    {
                        "audio": datasets.Audio(sampling_rate=22050),
                        "mel": datasets.Image(),
                        "label": datasets.features.ClassLabel(names=_NAMES["all"]),
                        "gender": datasets.features.ClassLabel(names=_NAMES["gender"]),
                        "singing_method": datasets.features.ClassLabel(
                            names=_NAMES["singing_method"]
                        ),
                    }
                )
                if self.config.name == "default"
                else (
                    datasets.Features(
                        {
                            "mel": datasets.Image(),
                            "cqt": datasets.Image(),
                            "chroma": datasets.Image(),
                            "label": datasets.features.ClassLabel(names=_NAMES["all"]),
                            "gender": datasets.features.ClassLabel(
                                names=_NAMES["gender"]
                            ),
                            "singing_method": datasets.features.ClassLabel(
                                names=_NAMES["singing_method"]
                            ),
                        }
                    )
                )
            ),
            supervised_keys=("mel", "label"),
            homepage=_HOMEPAGE,
            license="CC-BY-NC-ND",
            version="1.2.0",
            task_templates=[
                ImageClassification(
                    task="image-classification",
                    image_column="mel",
                    label_column="label",
                )
            ],
        )

    def _str2md5(self, original_string: str):
        md5_obj = hashlib.md5()
        md5_obj.update(original_string.encode("utf-8"))
        return md5_obj.hexdigest()

    def _split_generators(self, dl_manager):
        dataset = []
        if self.config.name == "default":
            files = {}
            audio_files = dl_manager.download_and_extract(_URLS["audio"])
            mel_files = dl_manager.download_and_extract(_URLS["mel"])
            for fpath in dl_manager.iter_files([audio_files]):
                fname: str = os.path.basename(fpath)
                if fname.endswith(".wav"):
                    cls = os.path.basename(os.path.dirname(fpath)) + "/"
                    item_id = self._str2md5(cls + fname.split(".wa")[0])
                    files[item_id] = {"audio": fpath}

            for fpath in dl_manager.iter_files([mel_files]):
                fname = os.path.basename(fpath)
                if fname.endswith(".jpg"):
                    cls = os.path.basename(os.path.dirname(fpath)) + "/"
                    item_id = self._str2md5(cls + fname.split(".jp")[0])
                    files[item_id]["mel"] = fpath

            dataset = list(files.values())

        else:
            data_files = dl_manager.download_and_extract(_URLS["eval"])
            for fpath in dl_manager.iter_files([data_files]):
                fname = os.path.basename(fpath)
                if "mel" in fpath and fname.endswith(".jpg"):
                    dataset.append(fpath)

        categories = {}
        for name in _NAMES["all"]:
            categories[name] = []

        for data in dataset:
            fpath = data["audio"] if self.config.name == "default" else data
            label = os.path.basename(os.path.dirname(fpath))
            categories[label].append(data)

        testset, validset, trainset = [], [], []
        for cls in categories:
            random.shuffle(categories[cls])
            count = len(categories[cls])
            p80 = int(count * 0.8)
            p90 = int(count * 0.9)
            trainset += categories[cls][:p80]
            validset += categories[cls][p80:p90]
            testset += categories[cls][p90:]

        random.shuffle(trainset)
        random.shuffle(validset)
        random.shuffle(testset)
        return [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN, gen_kwargs={"files": trainset}
            ),
            datasets.SplitGenerator(
                name=datasets.Split.VALIDATION, gen_kwargs={"files": validset}
            ),
            datasets.SplitGenerator(
                name=datasets.Split.TEST, gen_kwargs={"files": testset}
            ),
        ]

    def _generate_examples(self, files):
        if self.config.name == "default":
            for i, item in enumerate(files):
                label: str = os.path.basename(os.path.dirname(item["audio"]))
                yield i, {
                    "audio": item["audio"],
                    "mel": item["mel"],
                    "label": label,
                    "gender": ("male" if label.split("_")[0] == "m" else "female"),
                    "singing_method": (
                        "Bel_Canto" if label.split("_")[1] == "bel" else "Folk_Singing"
                    ),
                }

        else:
            for i, fpath in enumerate(files):
                label = os.path.basename(os.path.dirname(fpath))
                yield i, {
                    "mel": fpath,
                    "cqt": fpath.replace("mel", "cqt"),
                    "chroma": fpath.replace("mel", "chroma"),
                    "label": label,
                    "gender": ("male" if label.split("_")[0] == "m" else "female"),
                    "singing_method": (
                        "Bel_Canto" if label.split("_")[1] == "bel" else "Folk_Singing"
                    ),
                }