Merge pull request #8 from andweber/phonetisaurus_git

sphinx-stt: added support for current github version of phonetisaurus

Ty

plugins/stt/pocketsphinx-stt/g2p.py

@@ -15,16 +15,21 @@
                              r'input symbols table')
 
 
-def execute(executable, fst_model, input, is_file=False, nbest=None):
+def execute(executable, version, fst_model, input, is_file=False, nbest=None):
     logger = logging.getLogger(__name__)
 
     cmd = [executable,
-           '--model=%s' % fst_model,
-           '--input=%s' % input,
-           '--words']
-
-    if is_file:
-        cmd.append('--isfile')
+               '--model=%s' % fst_model]
+    if version <= 0.8:
+        cmd.append('--input=%s' % input)
+        cmd.append('--words')
+        if is_file:
+            cmd.append('--isfile')   
+    else:        
+        if is_file:
+            cmd.append('--wordlist=%s' % input)
+        else:
+            cmd.append('--word=%s' % input)
 
     if nbest is not None:
         cmd.extend(['--nbest=%d' % nbest])
@@ -74,7 +79,7 @@ def execute(executable, fst_model, input, is_file=False, nbest=None):
 
 
 class PhonetisaurusG2P(object):
-    def __init__(self, executable, fst_model,
+    def __init__(self, executable, version, fst_model,
                  fst_model_alphabet='arpabet',
                  nbest=None):
         self._logger = logging.getLogger(__name__)
@@ -88,6 +93,8 @@ def __init__(self, executable, fst_model,
         self._logger.debug("Using FST model alphabet: '%s'",
                            self.fst_model_alphabet)
 
+        self.version = version
+
         self.nbest = nbest
         if self.nbest is not None:
             self._logger.debug("Will use the %d best results.", self.nbest)
@@ -107,7 +114,7 @@ def _convert_phonemes(self, data):
         raise ValueError('Invalid FST model alphabet!')
 
     def _translate_word(self, word):
-        return execute(self.executable, self.fst_model, word,
+        return execute(self.executable, self.version, self.fst_model, word,
                        nbest=self.nbest)
 
     def _translate_words(self, words):
@@ -118,7 +125,7 @@ def _translate_words(self, words):
             for word in words:
                 f.write("%s\n" % word)
             tmp_fname = f.name
-        output = execute(self.executable, self.fst_model, tmp_fname,
+        output = execute(self.executable, self.version, self.fst_model, tmp_fname,
                          is_file=True, nbest=self.nbest)
         os.remove(tmp_fname)
         return output

plugins/stt/pocketsphinx-stt/sphinxvocab.py

@@ -44,6 +44,11 @@ def compile_vocabulary(config, directory, phrases):
     except KeyError:
         executable = 'phonetisaurus-g2p'
 
+    try:
+        version = config['pocketsphinx']['phonetisaurus_version']
+    except KeyError:
+        version = 0.7
+
     try:
         nbest = config['pocketsphinx']['nbest']
     except KeyError:
@@ -65,7 +70,7 @@ def compile_vocabulary(config, directory, phrases):
     if not os.path.exists(fst_model):
         raise OSError('FST model does not exist!')
 
-    g2pconverter = PhonetisaurusG2P(executable, fst_model,
+    g2pconverter = PhonetisaurusG2P(executable, version, fst_model,
                                     fst_model_alphabet=fst_model_alphabet,
                                     nbest=nbest)
 

plugins/stt/pocketsphinx-stt/tests/test_g2p.py

@@ -30,7 +30,7 @@ def poll(self):
 
 class TestPatchedG2P(unittest.TestCase):
     def setUp(self):
-        self.g2pconv = g2p.PhonetisaurusG2P('dummy_proc', 'dummy_fst_model',
+        self.g2pconv = g2p.PhonetisaurusG2P('dummy_proc', 0.7, 'dummy_fst_model',
                                             nbest=3)
 
     def testTranslateWord(self):