Source Code for Module saui_pr4.text2words

 1  #!/usr/bin/env python 
 2   
 3  # Copyright (c) 2007 Carnegie Mellon University. 
 4  # 
 5  # You may modify and redistribute this file under the same terms as 
 6  # the CMU Sphinx system.  See 
 7  # http://cmusphinx.sourceforge.net/html/LICENSE for more information. 
 8  # 
 9  # Briefly, don't remove the copyright.  Otherwise, do what you like. 
10   
11  """ 
12  Take a text file and extract words from it using Flite's text 
13  normalization, yielding a transcription suitable for language model 
14  training and a pronunciation dictionary. 
15   
16  This is part of project 4 in 05-631 Software Architecture for User 
17  Interfaces, Fall 2007. 
18  """ 
19   
20  __author__ = "David Huggins-Daines <dhuggins@cs.cmu.edu>" 
21   
22  import flite 
23  import getopt 
24  import sys 
25   
26 -def usage(): 
27      """Show usage of this module.""" 
28      sys.stderr.write("""Usage: text2words.py [-o DICTIONARY] < TEXT > WORDS""") 
29   
30 -class TextNorm(object): 
31 -    def __init__(self): 
32          self.vox = flite.load_voice() 
33          self.vox.set_post_synth_callback(self.get_words) 
34          self.dic = {} 
35   
36 -    def normalize(self, par): 
37          self.vox.text_to_speech(par, "none") 
38   
39 -    def get_words(self, utt): 
40          words = [] 
41          for w in utt['Word']: 
42              segs = [] 
43              for syl in w.as_rel('SylStructure'): 
44                  for seg in syl: 
45                      if seg.name == 'ax': 
46                          segs.append('AH') 
47                      elif seg.name == 'pau': 
48                          segs.append('SIL') 
49                      else: 
50                          segs.append(seg.name.upper()) 
51              if len(segs) > 0: 
52                  words.append(w.name) 
53                  if not w.name in self.dic: 
54                      self.dic[w.name] = " ".join(segs) 
55          print "<s>", " ".join(words), "</s>" 
56   
57 -def main(): 
58      try: 
59          opts, args = getopt.getopt(sys.argv[1:], 'o:', ['output=']) 
60      except getopt.GetoptError: 
61          usage() 
62          sys.exit(2) 
63      outfile = None 
64      for o, a in opts: 
65          if o in ('-o', '--output'): 
66              outfile = a 
67   
68      norm = TextNorm() 
69      par = [] 
70      for line in sys.stdin: 
71          line = line.strip() 
72          if line == "": 
73              norm.normalize(" ".join(par)) 
74              par = [] 
75          else: 
76              par.append(line) 
77      if outfile != None: 
78          outfh = open(outfile, "w") 
79          words = norm.dic.keys() 
80          words.sort() 
81          for w in words: 
82              outfh.write("%-30s %s\n" % (w, norm.dic[w])) 
83   
84  if __name__ == "__main__": 
85      main() 
86