german_corpus.py 文件源码

def __init__(self, base_directory: Path):
        super().__init__(
            corpus_name="german-speechdata-package-v2",
            base_directory=base_directory,
            base_source_url_or_directory="http://www.repository.voxforge1.org/downloads/de/",
            tar_gz_extension=".tar.gz",
            subdirectory_depth=1,
            umlaut_decoder=UmlautDecoder.none,
            training_test_split=TrainingTestSplit.by_directory(),
            tags_to_ignore=[],
            # exclude those 7 audio files because the first 2 are corrupt, the last 5 are empty:
            id_filter_regex=re.compile("(?!^2014-03-24-13-39-24_Kinect-RAW)"
                                       "(?!^2014-03-27-11-50-33_Kinect-RAW)"
                                       "(?!^2014-03-18-15-34-19_Realtek)"
                                       "(?!^2014-06-17-13-46-27_Kinect-RAW)"
                                       "(?!^2014-06-17-13-46-27_Realtek)"
                                       "(?!^2014-06-17-13-46-27_Samson)"
                                       "(?!^2014-06-17-13-46-27_Yamaha)"
                                       "(^.*$)"))