Skip to content


Subversion checkout URL

You can clone with HTTPS or Subversion.

Download ZIP
branch: master
Fetching contributors…

Cannot retrieve contributors at this time

256 lines (190 sloc) 8.906 kb
from dataset.timit import TIMIT
from experiments import utils
import sys
from scikits.talkbox import segment_axis
import numpy as np
def build_aa_dataset(in_samples, out_samples, shift, n_train=100, n_valid=10):
aa_seqs = np.load('/data/lisa/data/timit/readable/per_phone/wav_aa.npy')
mean = np.mean(np.hstack(aa_seqs))
std = np.std(np.hstack(aa_seqs))
print "mean:%f , std:%f"%(mean,std)
aa_max,aa_min = np.max(np.hstack(aa_seqs)), np.min(np.hstack(aa_seqs))
norm_seqs = np.asarray([(seq.astype('float32')-mean)/std \
for seq in aa_seqs])
# n_seq = norm_seqs.shape[0]
# n_train = n_seq*9/10
# train_aa_seqs = norm_seqs[:n_train]
# valid_aa_seqs = norm_seqs[n_train:]
# n_train = 100
# n_valid = 10
train_aa_seqs = norm_seqs[:n_train]
valid_aa_seqs = norm_seqs[n_train:n_train+n_valid]
print 'train sequences:', train_aa_seqs.shape[0]
print 'valid sequences:', valid_aa_seqs.shape[0]
frame_len = in_samples + out_samples
overlap = frame_len - shift
train_samples = []
valid_samples = []
for wav_seq in train_aa_seqs:
train_samples.append(segment_axis(wav_seq, frame_len, overlap))
train_samples = np.vstack(train_samples[:])
train_samples = np.random.permutation(train_samples)
for wav_seq in valid_aa_seqs:
valid_samples.append(segment_axis(wav_seq, frame_len, overlap))
valid_samples = np.vstack(valid_samples[:])
print 'train examples:', train_samples.shape
print 'valid examples:', valid_samples.shape
train_x = train_samples[:,:in_samples]
train_y = train_samples[:,in_samples:]
print train_x.shape, train_y.shape
valid_x = valid_samples[:,:in_samples]
valid_y = valid_samples[:,in_samples:]
print valid_x.shape, valid_y.shape
return utils.shared_dataset(train_x), \
utils.shared_dataset(train_y), \
utils.shared_dataset(valid_x), \
def build_one_user_data(dataset, in_samples, out_samples, shift,
win_width, shuffle, usr_id=0):
"""a function that builds train and validation set for one user
in the training set"""
print "building datasets for user %d"%usr_id
subset = 'train'
train_wav_seqs = dataset.train_raw_wav[usr_id*10:usr_id*10+9]
train_seqs_to_phns = dataset.train_seq_to_phn[usr_id*10:usr_id*10+9]
train_x, train_y1, train_y2 = \
_build_frames_w_phn(dataset, subset,
train_wav_seqs, train_seqs_to_phns,
in_samples, out_samples, shift,
win_width, shuffle)
valid_wav_seqs = dataset.train_raw_wav[usr_id*10+9:(usr_id+1)*10]
valid_seqs_to_phns = dataset.train_seq_to_phn[usr_id*10+9:(usr_id+1)*10]
#import pdb; pdb.set_trace()
valid_x, valid_y1, valid_y2 = \
_build_frames_w_phn(dataset, subset,
valid_wav_seqs, valid_seqs_to_phns,
in_samples, out_samples, shift,
win_width, shuffle)
return train_x, train_y1, train_y2, valid_x, valid_y1, valid_y2
def build_data_sets(dataset, subset, n_spkr, n_utts,
in_samples, out_samples, shift,
win_width, shuffle):
"""general function that builds data sets for training/validating/testing
the models from the corresponding dataset in TIMIT"""
print "building %s dataset..."%subset
wav_seqs = dataset.__dict__[subset+"_raw_wav"][0:n_utts*n_spkr]
seqs_to_phns = dataset.__dict__[subset+"_seq_to_phn"][0:n_utts*n_spkr]
return _build_frames_w_phn(dataset, subset, wav_seqs, seqs_to_phns,
in_samples, out_samples, shift,
win_width, shuffle)
def _build_frames_w_phn(dataset, subset, wav_seqs, seqs_to_phns,
in_samples, out_samples, shift,
win_width, shuffle):
#import pdb; pdb.set_trace()
norm_seqs = utils.standardize(wav_seqs)
#norm_seqs = utils.normalize(wav_seqs)
frame_len = in_samples + out_samples
overlap = frame_len - shift
samples = []
seqs_phn_info = []
seqs_phn_shift = []
# CAUTION!: I am using here reduced phone set
# we can also try using the full set but we must store phn+1
# because 0 no more refers to 'h#' (no speech)
for ind in range(len(norm_seqs)):
#import pdb; pdb.set_trace()
wav_seq = norm_seqs[ind]
phn_seq = seqs_to_phns[ind]
phn_start_end = dataset.__dict__[subset+"_phn"][phn_seq[0]:phn_seq[1]]
# create a matrix with consecutive windows
# phones are padded by h#, because each window will be shifted once
# the first phone samples has passed
phones = np.append(phn_start_end[:,2].astype('int16'),
# phones = np.append(phn_start_end[:,2],
# np.zeros((1,)))
phn_windows = segment_axis(phones, win_width, win_width-1)
# array that has endings of each phone
phn_ends = phn_start_end[:,1]
# extend the last phone till the end, this is not wrong as long as the
# last phone is no speech phone (h#)
phn_ends[-1] = wav_seq.shape[0]-1
# create a mapping from each sample to phn_window
phn_win_shift = np.zeros_like(wav_seq,dtype='int16')
phn_win_shift[phn_ends] = 1
phn_win = phn_win_shift.cumsum(dtype='int16')
# minor correction!
phn_win[-1] = phn_win[-2]
# Segment samples into frames
samples.append(segment_axis(wav_seq, frame_len, overlap))
# for phones we care only about one value to mark the start of a new window.
# the start of a phone window in a frame is when all samples of previous
# phone hav passed, so we use 'min' function to choose the current phone
# of the frame
phn_frames = segment_axis(phn_win, frame_len, overlap).min(axis=1)
# replace the window index with the window itself
win_frames = phn_windows[phn_frames]
#import pdb; pdb.set_trace()
# create a window shift for each frame
shift_frames_aux = np.roll(phn_frames,1)
shift_frames_aux[0] = 0
shift_frames = phn_frames - shift_frames_aux
# to mark the ending of the sequence - countering the first correction!
shift_frames[-1] = 1
#import pdb; pdb.set_trace()
#import pdb; pdb.set_trace()
# stack all data in one matrix, each row is a frame
samples_data = np.vstack(samples[:])
phn_data = np.vstack(seqs_phn_info[:])
shift_data = np.hstack(seqs_phn_shift[:])
#convert phone data to one-hot
from pylearn2.format.target_format import OneHotFormatter
fmt = OneHotFormatter(max_labels=39, dtype='float32')
phn_data = fmt.format(phn_data)
phn_data = phn_data.reshape(phn_data.shape[0],
full_data = np.hstack([samples_data[:,:in_samples], phn_data, #input
samples_data[:,in_samples:], #out1
shift_data.reshape(shift_data.shape[0],1)]) #out2
if shuffle:
full_data = np.random.permutation(full_data)
data_x = full_data[:,:in_samples+win_width*39]
data_y1 = full_data[:,in_samples+win_width*39:-1]
data_y2 = full_data[:,-1]
print 'Done'
print 'There are %d examples in %s set'%(data_x.shape[0],subset)
print "--------------"
print 'data_x.shape', data_x.shape
print 'data_y1.shape', data_y1.shape
return utils.shared_dataset(data_x), \
if __name__ == "__main__":
print 'loading data...'
save_stdout = sys.stdout
sys.stdout = open('timit.log', 'w')
# creating wrapper object for TIMIT dataset
dataset = TIMIT()
sys.stdout = save_stdout
in_samples = 240
out_samples = 1
shift = 1
win_width = 2
# n_spkr = 1
# n_utts = 10
shuffle = False
# each training example has 'in_sample' inputs and 'out_samples' output
# and examples are shifted by 'shift'
build_one_user_data(dataset, in_samples, out_samples, shift,
win_width, shuffle)
## code for loading AA data
# in_samples = 240
# out_samples = 1
# shift = 1
# build_aa_dataset(in_samples, out_samples, shift)
Jump to Line
Something went wrong with that request. Please try again.