def __init__(self, step=10, window=20, max_freq=8000, mfcc_dim=13,
minibatch_size=20, desc_file=None, spectrogram=True, max_duration=10.0,
sort_by_duration=False):
"""
Params:
step (int): Step size in milliseconds between windows (for spectrogram ONLY)
window (int): FFT window size in milliseconds (for spectrogram ONLY)
max_freq (int): Only FFT bins corresponding to frequencies between
[0, max_freq] are returned (for spectrogram ONLY)
desc_file (str, optional): Path to a JSON-line file that contains
labels and paths to the audio files. If this is None, then
load metadata right away
"""
self.feat_dim = calc_feat_dim(window, max_freq)
self.mfcc_dim = mfcc_dim
self.feats_mean = np.zeros((self.feat_dim,))
self.feats_std = np.ones((self.feat_dim,))
self.rng = random.Random(RNG_SEED)
if desc_file is not None:
self.load_metadata_from_desc_file(desc_file)
self.step = step
self.window = window
self.max_freq = max_freq
self.cur_train_index = 0
self.cur_valid_index = 0
self.cur_test_index = 0
self.max_duration=max_duration
self.minibatch_size = minibatch_size
self.spectrogram = spectrogram
self.sort_by_duration = sort_by_duration
def __init__(self, step=10, window=20, max_freq=8000, mfcc_dim=13,
minibatch_size=20, desc_file=None, spectrogram=True, max_duration=10.0,
sort_by_duration=False):
"""
Params:
step (int): Step size in milliseconds between windows (for spectrogram ONLY)
window (int): FFT window size in milliseconds (for spectrogram ONLY)
max_freq (int): Only FFT bins corresponding to frequencies between
[0, max_freq] are returned (for spectrogram ONLY)
desc_file (str, optional): Path to a JSON-line file that contains
labels and paths to the audio files. If this is None, then
load metadata right away
"""
self.feat_dim = calc_feat_dim(window, max_freq)
self.mfcc_dim = mfcc_dim
self.feats_mean = np.zeros((self.feat_dim,))
self.feats_std = np.ones((self.feat_dim,))
self.rng = random.Random(RNG_SEED)
if desc_file is not None:
self.load_metadata_from_desc_file(desc_file)
self.step = step
self.window = window
self.max_freq = max_freq
self.cur_train_index = 0
self.cur_valid_index = 0
self.cur_test_index = 0
self.max_duration=max_duration
self.minibatch_size = minibatch_size
self.spectrogram = spectrogram
self.sort_by_duration = sort_by_duration
def __init__(self,
step=10,
window=20,
max_freq=8000,
desc_file=None,
pad=0):
"""
Params:
step (int): Step size in milliseconds between windows
window (int): FFT window size in milliseconds
max_freq (int): Only FFT bins corresponding to frequencies between
[0, max_freq] are returned
desc_file (str, optional): Path to a JSON-line file that contains
labels and paths to the audio files. If this is None, then
load metadata right away
"""
self.feat_dim = calc_feat_dim(window, max_freq)
self.feats_mean = np.zeros((self.feat_dim, ))
self.feats_std = np.ones((self.feat_dim, ))
if desc_file is not None:
self.load_metadata_from_desc_file(desc_file)
self.step = step
self.window = window
self.max_freq = max_freq
self.pad = pad
def __init__(self,
step=10,
window=20,
max_freq=8000,
mfcc_dim=13,
minibatch_size=20,
desc_file=None,
spectrogram=True,
max_duration=10.0,
sort_by_duration=False):
self.feat_dim = calc_feat_dim(window, max_freq)
self.mfcc_dim = mfcc_dim
self.feats_mean = np.zeros((self.feat_dim, ))
self.feats_std = np.ones((self.feat_dim, ))
self.rng = random.Random(RNG_SEED)
if desc_file is not None:
self.load_metadata_from_desc_file(desc_file)
self.step = step
self.window = window
self.max_freq = max_freq
self.cur_train_index = 0
self.cur_valid_index = 0
self.cur_test_index = 0
self.max_duration = max_duration
self.minibatch_size = minibatch_size
self.spectrogram = spectrogram
self.sort_by_duration = sort_by_duration
