def generate_humavips_dataset(
glob_file_pattern="/mnt/protolab_server/media/sounds/datasets/NAR_dataset/*/*.wav"
):
dict_with_features = _generate_humavips_dataset(
glob_file_pattern=glob_file_pattern)
df = pd.DataFrame(dict_with_features)
df['features'] = df['features'].apply(lambda x: _flatten_features_dict(x))
return df
def generate_8k_dataset(glob_file_pattern='/mnt/protolab_server_8k/fold*/*.wav', nfft=1024, downsampling_freq=None):
dict_with_features = _generate_8k_dataset_dict(glob_file_pattern=glob_file_pattern, nfft=nfft)
df = pd.DataFrame(dict_with_features)
df['fold'] = df['file_path'].apply(lambda x: int(os.path.basename(x)[4:])) # string = foldXY , so we take string[4:
df['features'] = df['features'].apply(lambda x : _flatten_features_dict(x))
df['expected_class'] = df['file_name'].apply(lambda x: _add_class_from_filename_8kdataset(x))
return df
def generate_aldebaran_dataset(files, nfft=1024, window_block=None):
dict_with_features = _generate_aldebaran_dataset(files, window_block=window_block)
df = pd.DataFrame(dict_with_features)
df['features'] = df['features'].apply(lambda x : _flatten_features_dict(x))
return df
def generate_humavips_dataset(glob_file_pattern="/mnt/protolab_server/media/sounds/datasets/NAR_dataset/*/*.wav"):
dict_with_features = _generate_humavips_dataset(glob_file_pattern=glob_file_pattern)
df = pd.DataFrame(dict_with_features)
df['features'] = df['features'].apply(lambda x : _flatten_features_dict(x))
return df
def generate_aldebaran_dataset(files, nfft=1024, window_block=None):
dict_with_features = _generate_aldebaran_dataset(files,
window_block=window_block)
df = pd.DataFrame(dict_with_features)
df['features'] = df['features'].apply(lambda x: _flatten_features_dict(x))
return df
