-
Notifications
You must be signed in to change notification settings - Fork 1
/
scripts.py
286 lines (241 loc) · 9.56 KB
/
scripts.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
import glob
import os
from pylab import *
rcParams['figure.figsize'] = 8, 6
def groundtruths2mlf(groundtruthsdir, mlf):
def toHTKTime(second):
ten_million = 10000000
return int(second * ten_million)
lines = ["#!MLF!#"]
for filename in os.listdir(groundtruthsdir):
if filename.endswith('.txt'):
name = filename.split('.')[0]
print name
lines.append('"*/{0}.lab"'.format(name))
with open(groundtruthsdir + os.sep + filename, 'r') as f:
prev = -1.0 # seconds
for line in f:
things = line.split()
if len(things) == 3:
start = float(things[0])
end = float(things[1])
if start - prev < 0.20 and prev >= 0: # <200ms SP
lines.append('{0} {1} SP'.format(toHTKTime(prev), toHTKTime(start)))
lines.append('{0} {1} NOTE'.format(toHTKTime(start), toHTKTime(end)))
else: # >200ms SIL
if prev < 0:
prev = 0
lines.append('{0} {1} SIL'.format(toHTKTime(prev), toHTKTime(start)))
lines.append('{0} {1} NOTE'.format(toHTKTime(start), toHTKTime(end)))
prev = end
lines.append("SIL")
lines.append(".")
with open(mlf, "w") as g:
g.write("\n".join(lines)+"\n")
def groundtruths2mlf_separateNotes(groundtruthsdir, mlf, monophonesfile=None):
def toHTKTime(second):
ten_million = 10000000
return int(second * ten_million)
lines = ["#!MLF!#"]
note_count = 0
for filename in os.listdir(groundtruthsdir):
if filename.endswith('.txt'):
name = filename.split('.')[0]
print name
lines.append('"*/{0}.lab"'.format(name))
with open(groundtruthsdir + os.sep + filename, 'r') as f:
prev = -1.0 # seconds
for line in f:
things = line.split()
if len(things) == 3:
note_count += 1
start = float(things[0])
end = float(things[1])
if start - prev < 0.20 and prev >= 0: # <200ms SP
lines.append('{0} {1} SP'.format(toHTKTime(prev), toHTKTime(start)))
lines.append('{0} {1} NOTE_{2}'.format(toHTKTime(start), toHTKTime(end), note_count))
else: # >200ms SIL
if prev < 0:
prev = 0
lines.append('{0} {1} SIL'.format(toHTKTime(prev), toHTKTime(start)))
lines.append('{0} {1} NOTE_{2}'.format(toHTKTime(start), toHTKTime(end), note_count))
prev = end
lines.append("SIL")
lines.append(".")
with open(mlf, "w") as g:
g.write("\n".join(lines)+"\n")
if monophonesfile:
with open(monophonesfile, 'w') as g:
g.write('SIL\n')
g.write('SP\n')
for i in range(note_count):
g.write("{0}\n".format(i+1))
def plot_hmmdefs(hmmdef_file):
with open(hmmdef_file) as f:
lines = f.read().split('\n')
print lines[3]
assert lines[3].strip() == '~h "UNVOICED"'
print lines[31]
assert lines[31].strip() == '~h "VOICED"'
u2ms = map(float32, lines[8].strip().split())
u2vs = map(float32, lines[10].strip().split())
u3ms = map(float32, lines[14].strip().split())
u3vs = map(float32, lines[16].strip().split())
u4ms = map(float32, lines[20].strip().split())
u4vs = map(float32, lines[22].strip().split())
v2ms = map(float32, lines[36].strip().split())
v2vs = map(float32, lines[38].strip().split())
v3ms = map(float32, lines[42].strip().split())
v3vs = map(float32, lines[44].strip().split())
v4ms = map(float32, lines[48].strip().split())
v4vs = map(float32, lines[50].strip().split())
print u2ms, u2vs, v2ms, v2vs
titles = ["Pitch Difference", "Energy", "Voicing", "Accent"]
for i in range(4):
subplot(4, 1, i+1)
xlpos = [1, 2, 3, 5, 6, 7]
xlabels = ["voiced-1", "voiced-2", "voiced-3", "unvoiced-1", "unvoiced-2", "unvoiced-3"]
errorbar(xlpos,
[v2ms[i], v3ms[i], v4ms[i], u2ms[i], u3ms[i], u4ms[i]],
[v2vs[i], v3vs[i], v4vs[i], u2vs[i], u3vs[i], u4vs[i]],
linestyle="None", marker="^")
xticks(xlpos, xlabels, size="small")
xlim((xlpos[0]-1, xlpos[-1]+1))
if i == 2: # voicing
ylim((0, 1))
elif i == 3: # accent
ylim((0, 80))
elif i == 0: # pitch diff
ylim((-1e10, 1e9))
title(titles[i])
grid(axis="y")
show()
if False:
i = 0
xlpos = [1, 2, 3, 5, 6, 7]
xlabels = ["voiced-1", "voiced-2", "voiced-3", "unvoiced-1", "unvoiced-2", "unvoiced-3"]
errorbar(xlpos,
[v2ms[i], v3ms[i], v4ms[i], u2ms[i], u3ms[i], u4ms[i]],
[v2vs[i], v3vs[i], v4vs[i], u2vs[i], u3vs[i], u4vs[i]],
linestyle="None", marker="^")
xticks(xlpos, xlabels, size="small")
xlim((xlpos[0]-1, xlpos[-1]+1))
ylim((0, 15))
title(titles[i])
grid(axis="y")
show()
def split_test_mlf(mlf_file, test_dir, ext):
with open(mlf_file) as f:
content = f.read()
files = content.split('.\n')
for f in files:
f = f.strip()
lines = f.split('\n')
if lines[0] == "#!MLF!#":
lines = lines[1:]
basename = lines[0][3:-5]
print basename
contents = []
for line in lines[1:]:
if len(line.split()) >= 3:
_, _, is_voiced = line.split()[:3]
if is_voiced == 'NOTE':
contents.append(line)
def transform(line):
start, end = line.split()[:2]
start = str(float(start) / 10000000.0)
end = str(float(end) / 10000000.0)
return start + ' ' + end + ' 1'
contents = map(transform, contents)
with open(test_dir + os.sep + basename + '.' + ext, 'w') as g:
g.write('\n'.join(contents)+'\n')
def tony_format_to_evaluation_framework(tony_format_dir, output_dir, ext):
for filename in os.listdir(tony_format_dir):
basename = filename.split('.')[0]
print basename
with open(tony_format_dir + os.sep + filename, 'r') as f:
with open(output_dir + os.sep + basename + "." + ext, 'w') as g:
for line in f:
start, pit, dur = line.split()[:3]
g.write("{0} {1} {2}\n".format(start, float(start)+float(dur), 1))
def extract_notes_from_onsets(labfile, pitfile, outfile):
print labfile
notes = []
with open(labfile) as f:
for line in f:
notes.append(line.strip().split())
fs = 44100
hopsize = 512
pits = [] # hopsize = 512
with open(pitfile) as f:
pits = map(float32, f.read().split())
ten_million = 10000000
extracted_notes = []
for note in notes:
if note[2] == "VOICED":
start = float32(note[0]) / ten_million
end = float32(note[1]) / ten_million
start_b = floor(start*fs/hopsize)
end_b = floor(end*fs/hopsize)
candidates = []
for i in range(int(start_b), int(end_b)+1):
p = pits[i]
if p > 70 and p < 880:
candidates.append(p)
if not candidates:
print "WARNING:: ignoring this note"
continue
out_p = median(candidates)
print out_p
extracted_notes.append("{0} {1} {2}".format(start, end, aubio.freqtomidi(out_p)))
with open(outfile, 'w') as g:
g.write('\n'.join(extracted_notes)+'\n')
def pitch_to_1_ground_truths(src, target):
with open(src) as f:
with open(target, 'w') as g:
for line in f:
start, end = line.split()[:2]
g.write(start+' '+end+' 1\n')
def analyze_GArun(GArun_dir):
# read genes and results
genes_map = {}
results_map = {}
results_list = []
len_gene = None
with open(GArun_dir+os.sep+"genes") as f:
for line in f:
i, g = line.split()
genes_map[i] = map(int, list(g))
len_gene = len(g)
with open(GArun_dir+os.sep+"results") as f:
for line in f:
i, r = line.split()
results_map[i] = float(r)
results_list.append((i, float(r)))
sorted_results_list = sorted(results_list, key=lambda a: a[1], reverse=True)
sums = [0] * len_gene
for i, r in sorted_results_list[:30]:
gene = genes_map[i]
sums = map(sum, zip(sums, gene))
print sums
bar(range(len_gene), sums)
show()
def GA_evolution(*dirs):
maxs = [0.0]*101
avgs = [0.0]*101
for d in dirs:
with open(d+os.sep+"evolution.txt") as f:
for line in f:
g, a, m = line.split()[:3]
g = int(g)
a = float(a)
m = float(m)
avgs[g] += a / len(dirs)
maxs[g] += m / len(dirs)
plot(range(101), avgs, marker="x", label='Average')
plot(range(101), maxs, marker="o", label='Best')
xlabel('Generation #')
ylabel('Correct Onset & Offset (F-measure)')
legend(loc=2)
#ylim((0, 1.0))
show()