def test_compile_features_for_word(self):
context = {-2: set([("foo", 1), ("bar", 1), ("dur", 0.377)]), -1: set([("fii", 1), ("dur", 0.553)]), +1: set([("</s>", 1)])}
local_feature_and_dur_seq = [(set([("bii", 1), ("boo", 1)]), 5), (set([("buu", 1), ("bee", 1)]), 6)]
result = durmodel_utils.compile_features_for_word(context, local_feature_and_dur_seq)
#print result[0]
self.assertEqual(result[0][1], 5)
self.assert_(("pos-2:foo", 1) in result[0][0])
self.assert_(("bii", 1) in result[0][0])
self.assert_(("pos-1:dur", 0.553) in result[0][0])
self.assert_(("pos-2:fii", 1) in result[1][0])
self.assert_(("pos+2:</s>", 1) in result[-2][0])
self.assert_(("pos+1:</s>", 1) in result[-1][0])
self.assert_(("pos-1:dur", durmodel_utils.dur_function(5)) in result[-1][0])
def get_X_raw_utt(utt_features_and_durs, feature_dict):
X_raw_utt = np.zeros(
(len(utt_features_and_durs), len(feature_dict)), dtype=np.float16)
prev_duration = 5
for i, (_features, _dur) in enumerate(utt_features_and_durs):
X_raw_utt[i, 0] = durmodel_utils.dur_function(prev_duration)
for feature_name, value in _features:
feature_id = feature_dict.get(feature_name, -1)
if feature_id >= 0:
X_raw_utt[i, feature_id] = value
prev_duration = _dur
return X_raw_utt
def _test_linear_lattice(self):
seqs = []
words = ['<eps>', 'komm', 'Tanel']
# komm
seqs.append((
1,
'59_59_59_59_59_93_93_93_93_93_93_93_80_80_80_80_80_80_80_80_80_80_80_80_80_80_80_80_80_80_80'
.split('_')))
# SIL
seqs.append((0, '6_6_6_6_6_6_6'.split('_')))
# Tanel
seqs.append((
2,
'135_135_135_135_135_135_135_135_13_13_13_13_13_13_13_13_13_13_85_85_85_85_85_85_85_25_25_25_25_25_25_25_25_25_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68'
.split('_')))
phone_map = {
6: 'SIL',
13: 'a',
25: 'e',
59: 'k',
68: 'l',
93: 'o',
80: 'mm',
85: 'n',
135: 't'
}
features_and_durs = []
for (word, seq) in seqs:
features_and_durs.append(
durmodel_utils.make_local(word, seq, phone_map, words))
# print features_and_durs
full_features_and_durs = durmodel_utils.make_linear(features_and_durs)
self.assertEqual(8, len(full_features_and_durs))
print(full_features_and_durs[1][0])
self.assert_(
'pos-1:dur=%f' %
durmodel_utils.dur_function(5) in full_features_and_durs[1][0])
for (f, d) in full_features_and_durs:
print('%d %s' % (d, ' '.join(f)))
def _test_linear_lattice(self):
seqs = []
words = ["<eps>", "komm", "Tanel"]
#komm
seqs.append((1, "59_59_59_59_59_93_93_93_93_93_93_93_80_80_80_80_80_80_80_80_80_80_80_80_80_80_80_80_80_80_80".split("_")))
#SIL
seqs.append((0, "6_6_6_6_6_6_6".split("_")))
#Tanel
seqs.append((2, "135_135_135_135_135_135_135_135_13_13_13_13_13_13_13_13_13_13_85_85_85_85_85_85_85_25_25_25_25_25_25_25_25_25_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68".split("_")))
phone_map = {6: "SIL", 13: "a", 25: "e", 59 : "k", 68: "l", 93 : "o", 80: "mm", 85 : "n", 135 : "t"}
features_and_durs = []
for (word, seq) in seqs:
features_and_durs.append(durmodel_utils.make_local(word, seq, phone_map, words))
#print features_and_durs
full_features_and_durs = durmodel_utils.make_linear(features_and_durs)
self.assertEqual(8, len(full_features_and_durs))
print full_features_and_durs[1][0]
self.assert_("pos-1:dur=%f" % durmodel_utils.dur_function(5) in full_features_and_durs[1][0])
for (f, d) in full_features_and_durs:
print "%d %s" % (d, " ".join(f))
def test_compile_features_for_word(self):
context = {
-2: set([('foo', 1), ('bar', 1), ('dur', 0.377)]),
-1: set([('fii', 1), ('dur', 0.553)]),
+1: set([('</s>', 1)])
}
local_feature_and_dur_seq = [(set([('bii', 1), ('boo', 1)]), 5),
(set([('buu', 1), ('bee', 1)]), 6)]
result = durmodel_utils.compile_features_for_word(
context, local_feature_and_dur_seq)
# print result[0]
self.assertEqual(result[0][1], 5)
self.assert_(('pos-2:foo', 1) in result[0][0])
self.assert_(('bii', 1) in result[0][0])
self.assert_(('pos-1:dur', 0.553) in result[0][0])
self.assert_(('pos-2:fii', 1) in result[1][0])
self.assert_(('pos+2:</s>', 1) in result[-2][0])
self.assert_(('pos+1:</s>', 1) in result[-1][0])
self.assert_(('pos-1:dur',
durmodel_utils.dur_function(5)) in result[-1][0])
def test_compile_features_for_word(self):
context = {
-2: set([("foo", 1), ("bar", 1), ("dur", 0.377)]),
-1: set([("fii", 1), ("dur", 0.553)]),
+1: set([("</s>", 1)])
}
local_feature_and_dur_seq = [(set([("bii", 1), ("boo", 1)]), 5),
(set([("buu", 1), ("bee", 1)]), 6)]
result = durmodel_utils.compile_features_for_word(
context, local_feature_and_dur_seq)
#print result[0]
self.assertEqual(result[0][1], 5)
self.assert_(("pos-2:foo", 1) in result[0][0])
self.assert_(("bii", 1) in result[0][0])
self.assert_(("pos-1:dur", 0.553) in result[0][0])
self.assert_(("pos-2:fii", 1) in result[1][0])
self.assert_(("pos+2:</s>", 1) in result[-2][0])
self.assert_(("pos+1:</s>", 1) in result[-1][0])
self.assert_(("pos-1:dur",
durmodel_utils.dur_function(5)) in result[-1][0])
def _test_linear_lattice(self):
seqs = []
words = ["<eps>", "komm", "Tanel"]
#komm
seqs.append((
1,
"59_59_59_59_59_93_93_93_93_93_93_93_80_80_80_80_80_80_80_80_80_80_80_80_80_80_80_80_80_80_80"
.split("_")))
#SIL
seqs.append((0, "6_6_6_6_6_6_6".split("_")))
#Tanel
seqs.append((
2,
"135_135_135_135_135_135_135_135_13_13_13_13_13_13_13_13_13_13_85_85_85_85_85_85_85_25_25_25_25_25_25_25_25_25_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68_68"
.split("_")))
phone_map = {
6: "SIL",
13: "a",
25: "e",
59: "k",
68: "l",
93: "o",
80: "mm",
85: "n",
135: "t"
}
features_and_durs = []
for (word, seq) in seqs:
features_and_durs.append(
durmodel_utils.make_local(word, seq, phone_map, words))
#print features_and_durs
full_features_and_durs = durmodel_utils.make_linear(features_and_durs)
self.assertEqual(8, len(full_features_and_durs))
print full_features_and_durs[1][0]
self.assert_(
"pos-1:dur=%f" %
durmodel_utils.dur_function(5) in full_features_and_durs[1][0])
for (f, d) in full_features_and_durs:
print "%d %s" % (d, " ".join(f))
with codecs.open(args.write_features_filename,
'w',
encoding="UTF-8") as f:
for feature in feature_dict:
print >> f, feature
raw_data_X = []
for utt_features_and_durs in all_features_and_durs:
utt_raw_data_X = np.zeros(
(len(utt_features_and_durs), len(feature_dict)),
dtype=np.float16)
prev_duration = 5
for (i, (_features, _dur)) in enumerate(utt_features_and_durs):
utt_raw_data_X[i,
0] = durmodel_utils.dur_function(prev_duration)
for feature in _features:
(feature_name, value) = feature
feature_id = feature_dict.get(feature_name, -1)
if feature_id >= 0:
utt_raw_data_X[i, feature_id] = value
prev_duration = _dur
raw_data_X.append(utt_raw_data_X)
raw_data_y = []
for utt_features_and_durs in all_features_and_durs:
utt_raw_data_y = np.zeros((len(utt_features_and_durs), 2),
dtype=np.int16)
utt_raw_data_y[:, 0] = np.array(
[d for (_, d) in utt_features_and_durs])
