train_get_1_best.py
3.09 KB
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
#!/usr/bin/env python
# Copyright 2014 Gaurav Kumar. Apache 2.0
import os
import sys
files = [
open('/export/a04/gkumar/kaldi-trunk/egs/fishcall_es/j-1/exp/tri5a/decode_test/scoring/13.tra'),
open('/export/a04/gkumar/kaldi-trunk/egs/fishcall_es/j-2/exp/tri5a/decode_test/scoring/13.tra'),
open('/export/a04/gkumar/kaldi-trunk/egs/fishcall_es/j-3/exp/tri5a/decode_test/scoring/13.tra'),
open('/export/a04/gkumar/kaldi-trunk/egs/fishcall_es/j-4/exp/tri5a/decode_test/scoring/13.tra'),
open('/export/a04/gkumar/kaldi-trunk/egs/fishcall_es/j-5/exp/tri5a/decode_test/scoring/13.tra'),
open('/export/a04/gkumar/kaldi-trunk/egs/fishcall_es/j-6/exp/tri5a/decode_test/scoring/13.tra'),
open('/export/a04/gkumar/kaldi-trunk/egs/fishcall_es/j-7/exp/tri5a/decode_test/scoring/13.tra'),
open('/export/a04/gkumar/kaldi-trunk/egs/fishcall_es/j-8/exp/tri5a/decode_test/scoring/13.tra'),
open('/export/a04/gkumar/kaldi-trunk/egs/fishcall_es/j-9/exp/tri5a/decode_test/scoring/13.tra'),
open('/export/a04/gkumar/kaldi-trunk/egs/fishcall_es/j-10/exp/tri5a/decode_test/scoring/13.tra')]
def findTranscription(timeDetail):
for file1 in files:
file1.seek(0,0)
for line in file1:
lineComp = line.split()
if lineComp[0] == timeDetail:
return " ".join(lineComp[1:])
# No result found
return -1
wordsFile = open('exp/tri5a/graph/words.txt')
words = {}
# Extract word list
for line in wordsFile:
lineComp = line.split()
words[int(lineComp[1])] = lineComp[0].strip()
# Now read list of files in conversations
fileList = []
#conversationList = open('/export/a04/gkumar/corpora/fishcall/joshkal-splits/provisional_dev')
conversationList = open('/export/a04/gkumar/corpora/fishcall/jack-splits/split-matt/train')
for line in conversationList:
line = line.strip()
line = line[:-4]
fileList.append(line)
# IN what order were the conversations added to the spanish files?
# TODO: Make sure they match the order in which these english files are being written
# Now get timing information to concatenate the ASR outputs
if not os.path.exists('exp/tri5a/one-best/train'):
os.makedirs('exp/tri5a/one-best/train')
#provFile = open('/export/a04/gkumar/corpora/fishcall/fisher_provisional_dev.es', 'w+')
provFile = open('/export/a04/gkumar/corpora/fishcall/jack-splits/split-matt/asr.train', 'w+')
for item in fileList:
timingFile = open('/export/a04/gkumar/corpora/fishcall/fisher/tim/' + item + '.es')
newFile = open('exp/tri5a/one-best/train/' + item + '.es', 'w+')
for line in timingFile:
timeInfo = line.split()
mergedTranslation = ""
for timeDetail in timeInfo:
#Locate this in ASR dev/test, this is going to be very slow
tmp = findTranscription(timeDetail)
if tmp != -1:
mergedTranslation = mergedTranslation + " " + tmp
mergedTranslation = mergedTranslation.strip()
transWords = [words[int(x)] for x in mergedTranslation.split()]
newFile.write(" ".join(transWords) + "\n")
provFile.write(" ".join(transWords) + "\n")
newFile.close()
provFile.close()