def test_aligners(japanese, english):
for i in range(len(english)):
out1 = aligner1.align_words(japanese[i],english[i])
out2 = aligner2.align_words(japanese[i],english[i], False)
if (out1 != out2):
if (out2 == "ERR"):
print "Basic: " + str(out1)
print "Ted : " + str(out2)
print "-"
def align_phoneme_sets(japanese,english,verbose=False):
"""
INPUT:
japanese - an array of japanese phoneme sequences. each item is a string
english - an array of english phoneme sequences. each item is a string
OUTPUT:
j_aligned - an array of japanese phoneme sequences. each item is an ARRAY
e_aligned - an array of english phoneme sequences. each item is an ARRAY
"""
import align_ted as aligner
print "[Start] Aligning Phoneme Sequences"
print "[.....] Starting with %i sequences" % len(japanese)
aligned_japanese = []
aligned_english = []
errors = 0
err_align = open("error_align.txt",'w')
for i in range(len(japanese)):
try:
jap_phones = japanese[i]
eng_phones = english[i]
aligned = aligner.align_words(jap_phones,eng_phones,False)
if (aligned == "ERR"):
err_align.write("line No:%d"%i+"\n")
err_align.write("Eng:%s"%jap_phones[i])
err_align.write("Jap:%s"%eng_phones[i]+"\n")
errors = errors + 1
else:
aligned_japanese.append(aligned[0])
aligned_english.append(aligned[1])
except:
if verbose:
print "Unexpected error line %i: %s" % (i, sys.exc_info()[0])
errors = errors + 1
err_align.close()
print "[.....] Success: %i, Error: %i, Error Rate: %f" % (len(aligned_japanese), errors, float(errors)/len(aligned_japanese))
print "[ End ] Aligning Phoneme Sequences"
return aligned_japanese, aligned_english
