phonemizer-2.0.1

bootphon · Nov 7, 2019 · 5ed5ca5 · 5ed5ca5
1 parent 6372005
commit 5ed5ca5
Show file tree

Hide file tree

Showing 2 changed files with 27 additions and 29 deletions.
diff --git a/phonemizer/backend/espeak.py b/phonemizer/backend/espeak.py
@@ -164,38 +164,37 @@ def _phonemize_aux(self, text, separator, strip):
                     if self.logger:
                         self.logger.debug('running %s', command)
 
-                    raw_output = subprocess.check_output(
+                    line = subprocess.check_output(
                         shlex.split(command, posix=False)).decode('utf8')
-
-                    # espeak can split an utterance into several lines because
-                    # of punctuation, here we merge the lines into a single one
-                    raw_output = raw_output.strip().replace('\n', ' ')
                 finally:
                     os.remove(data.name)
 
-                for line in (l.strip() for l in raw_output.split('\n') if l):
-                    line = self._process_lang_switch(n, line)
-                    if not line:
-                        continue
-
-                    out_line = ''
-                    for word in line.split(u' '):
-                        w = word.strip()
-
-                        # remove the stresses on phonemes
-                        if not self._with_stress:
-                            w = w.replace(u"ˈ", u'')
-                            w = w.replace(u'ˌ', u'')
-                            w = w.replace(u"'", u'')
-
-                        if not strip:
-                            w += '_'
-                        w = w.replace('_', separator.phone)
-                        out_line += w + separator.word
-
-                    if strip:
-                        out_line = out_line[:-len(separator.word)]
-                    output.append(out_line)
+                # espeak can split an utterance into several lines because
+                # of punctuation, here we merge the lines into a single one
+                line = line.strip().replace('\n', ' ').replace('  ', ' ')
+
+                line = self._process_lang_switch(n, line)
+                if not line:
+                    continue
+
+                out_line = ''
+                for word in line.split(u' '):
+                    w = word.strip()
+
+                    # remove the stresses on phonemes
+                    if not self._with_stress:
+                        w = w.replace(u"ˈ", u'')
+                        w = w.replace(u'ˌ', u'')
+                        w = w.replace(u"'", u'')
+
+                    if not strip:
+                        w += '_'
+                    w = w.replace('_', separator.phone)
+                    out_line += w + separator.word
+
+                if strip:
+                    out_line = out_line[:-len(separator.word)]
+                output.append(out_line)
 
         # warn the user on language switches fount during phonemization
         if self._lang_switch_list:

diff --git a/test/test_espeak.py b/test/test_espeak.py
@@ -130,7 +130,6 @@ def test_language_switch():
         'a comma. a point.',
         'a comma,, a point.',
         'a comma, , a point.',
-        'a comma, a point,',
         'a comma? a point!']
      for s in (True, False)
      for u in (separator.Separator(), separator.Separator(word='_', phone=' '))