erhu_playing_tech.py

import os
import random
import hashlib
import datasets


_NAMES = {
    "4_classes": [
        "trill",
        "staccato",
        "slide",
        "others",
    ],
    "7_classes": [
        "trill_short_up",
        "trill_long",
        "staccato",
        "slide_up",
        "slide_legato",
        "slide_down",
        "others",
    ],
    "11_classes": [
        "vibrato",
        "trill",
        "tremolo",
        "staccato",
        "ricochet",
        "pizzicato",
        "percussive",
        "legato_slide_glissando",
        "harmonic",
        "diangong",
        "detache",
    ],
}

_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 erhu_playing_tech(datasets.GeneratorBasedBuilder):
    def _info(self):
        if self.config.name == "default":
            self.config.name = "11_classes"

        return datasets.DatasetInfo(
            features=(
                datasets.Features(
                    {
                        "audio": datasets.Audio(sampling_rate=44100),
                        "mel": datasets.Image(),
                        "label": datasets.features.ClassLabel(
                            names=_NAMES[self.config.name]
                        ),
                    }
                )
                if self.config.name != "eval"
                else datasets.Features(
                    {
                        "mel": datasets.Image(),
                        "cqt": datasets.Image(),
                        "chroma": datasets.Image(),
                        "label": datasets.features.ClassLabel(
                            names=_NAMES["11_classes"]
                        ),
                    }
                )
            ),
            homepage=_HOMEPAGE,
            license="CC-BY-NC-ND",
            version="1.2.0",
        )

    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):
        if self.config.name != "eval":
            audio_files = dl_manager.download_and_extract(_URLS["audio"])
            mel_files = dl_manager.download_and_extract(_URLS["mel"])
            files = {}
            for fpath in dl_manager.iter_files([audio_files]):
                fname = os.path.basename(fpath)
                dirname = os.path.dirname(fpath)
                subset = os.path.basename(os.path.dirname(dirname))
                if self.config.name == subset and fname.endswith(".wav"):
                    cls = f"{subset}/{os.path.basename(dirname)}/"
                    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)
                dirname = os.path.dirname(fpath)
                subset = os.path.basename(os.path.dirname(dirname))
                if self.config.name == subset and fname.endswith(".jpg"):
                    cls = f"{subset}/{os.path.basename(dirname)}/"
                    item_id = self._str2md5(cls + fname.split(".jp")[0])
                    files[item_id]["mel"] = fpath

            dataset = list(files.values())

        else:
            eval_files = dl_manager.download_and_extract(_URLS["eval"])
            dataset = []
            for fpath in dl_manager.iter_files([eval_files]):
                fname: str = os.path.basename(fpath)
                if "_mel" in fname and fname.endswith(".jpg"):
                    dataset.append({"mel": fpath, "label": fname.split("__")[0]})

        categories = {}
        names = _NAMES["11_classes" if "eval" in self.config.name else self.config.name]
        for name in names:
            categories[name] = []

        for data in dataset:
            if self.config.name != "eval":
                data["label"] = os.path.basename(os.path.dirname(data["audio"]))

            categories[data["label"]].append(data)

        testset, validset, trainset = [], [], []
        for cls in categories:
            random.shuffle(categories[cls])
            count = len(categories[cls])
            p60 = int(count * 0.6)
            p80 = int(count * 0.8)
            trainset += categories[cls][:p60]
            validset += categories[cls][p60:p80]
            testset += categories[cls][p80:]

        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 != "eval":
            for i, item in enumerate(files):
                yield i, item

        else:
            for i, item in enumerate(files):
                yield i, {
                    "mel": item["mel"],
                    "cqt": item["mel"].replace("_mel", "_cqt"),
                    "chroma": item["mel"].replace("_mel", "_chroma"),
                    "label": item["label"],
                }