update scripts and merge

[asr1617data.git] / segment.py

diff --git a/segment.py b/segment.py
deleted file mode 100644 (file)
index dc803ef..0000000
--- a/segment.py
+++ /dev/null
@@ -1,60 +0,0 @@
-import sys
-import pympi
-import random
-import glob
-import re
-import os
-
-testset = 0.10
-data = []
-
-
-def process(num):
-    num = re.match('^.*/(\\d+).TextGrid$', num).group(1)
-    tg = 'textgrid/{:02d}.TextGrid'.format(int(num))
-
-    tgob = pympi.TextGrid(tg)
-    intervalit = tgob.get_tier('lyrics').get_intervals(sort=True)
-
-    with open('mfcc/{:02d}.mfcc'.format(int(num))) as mfcc:
-        (s, e, v) = next(intervalit)
-        currentframe = 0.0
-        for l in mfcc:
-            # Go to next interval
-            if currentframe > e:
-                try:
-                    (s, e, v) = next(intervalit)
-                except StopIteration:
-                    pass
-
-            label = 1 if v == '' else 0
-            data.append([label] + l.split('\t'))
-
-            # Increase time
-            currentframe += 0.01
-
-
-if __name__ == '__main__':
-    datafiles = []
-    for fl in glob.glob(os.path.join(os.getcwd(), 'textgrid', '*.TextGrid')):
-        sys.stderr.write('Segment {}\n'.format(fl))
-        process(fl)
-
-    sys.stderr.write('Shuffling {} samples\n'.format(len(data)))
-    random.shuffle(data)
-
-    splitpoint = int(len(data)*testset)
-    testset = data[:splitpoint]
-    trainset = data[splitpoint:]
-    del(data)
-
-    sys.stderr.write('Write testset: {} items\n'.format(splitpoint))
-    with open('test.txt', 'w') as f:
-        for d in testset:
-            f.write('\t'.join(map(str, d)))
-
-    sys.stderr.write('Write trainingset: {:d} items\n'.format(9*splitpoint))
-    with open('train.txt', 'w') as f:
-        for d in trainset:
-            f.write('\t'.join(map(str, d)))
-    f.close()