add first disgorge song
[asr1617data.git] / segment.py
index 3a7967f..dc803ef 100644 (file)
@@ -1,12 +1,16 @@
 import sys
 import pympi
 import random
+import glob
+import re
+import os
 
 testset = 0.10
 data = []
 
 
-def process(num, *args):
+def process(num):
+    num = re.match('^.*/(\\d+).TextGrid$', num).group(1)
     tg = 'textgrid/{:02d}.TextGrid'.format(int(num))
 
     tgob = pympi.TextGrid(tg)
@@ -32,11 +36,9 @@ def process(num, *args):
 
 if __name__ == '__main__':
     datafiles = []
-    with open('data.txt', 'r') as f:
-        for l in f:
-            s = l.strip().split('\t')
-            sys.stderr.write('Processing {}: {}\n'.format(s[0], s[1]))
-            process(*s)
+    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)