Merge branch 'master' of ssh://invmac/~/Public/repos/speech-scoring

tts_samplegen.py

@@ -11,7 +11,7 @@ import progressbar
 
 from generate_similar import similar_phoneme,similar_word
 
-OUTPUT_NAME = 'story_words'
+OUTPUT_NAME = 'rand_edu'
 dest_dir = os.path.abspath('.') + '/outputs/' + OUTPUT_NAME + '/'
 dest_file = './outputs/' + OUTPUT_NAME + '.csv'
 
@@ -95,8 +95,7 @@ class SynthVariant(object):
         self.name = voice
         self.lang = lang
         self.phoneme_capable = self.is_phoneme_capable()
-        if self.phoneme_capable:
-            create_dir(dest_dir + self.name + '/' + str(self.rate))
+
 
     def __repr__(self):
         return 'Synthesizer[{} - {}]'.format(self.name, self.rate)
@@ -128,6 +127,10 @@ class SynthVariant(object):
         cli_gen_audio(phon_cmd, self.rate, self.name, d_path)
         return SynthFile(word, phoneme, r_path, self.name, self.lang, self.rate, variant)
 
+    def create_synth_dirs(self):
+        if self.phoneme_capable:
+            create_dir(dest_dir + self.name + '/' + str(self.rate))
+
     @staticmethod
     def voices_for_lang(lang):
         voices_installed = NSSpeechSynthesizer.availableVoices()
@@ -168,9 +171,10 @@ def synth_generator():
 
     def synth_for_words(words, writer):
         prog_title = "Synthesizing {} words : ".format(len(words))
-        (update, prog) = prog_bar(prog_title)
         for s in voice_synths:
+            s.create_synth_dirs()
             for v in ['low', 'medium', 'high']:
+                (update, prog) = prog_bar(prog_title)
                 for w in prog(words):
                     update('"{}" with {} variant ({})'.format(w, s, v))
                     synthed = s.generate_audio(w, v)
@@ -209,13 +213,24 @@ def synth_logger(fname, csv=False):
     else:
         return json_writer, close_file
 
+def generate_audio_for_text_list(text_list):
+    (writer, closer) = synth_logger(dest_file, csv=True)
+    synth_for_words = synth_generator()
+    try:
+        synth_for_words(text_list, writer)
+    except:
+        import traceback
+        import sys
+        traceback.print_exc(file=sys.stdout)
+        pass
+    closer()
 
 def generate_audio_for_stories():
-    story_file = './inputs/all_stories_hs.json'
-    # story_file = './inputs/all_stories.json'
+    # story_file = './inputs/all_stories_hs.json'
+    story_file = './inputs/all_stories.json'
     stories_data = json.load(open(story_file))
-    word_list = [t[0] for i in stories_data.values() for t in i]
-    # word_list = [i for g in stories_data.values() for i in g]
+    # word_list = [t[0] for i in stories_data.values() for t in i]
+    word_list = [i for g in stories_data.values() for i in g]
     (writer, closer) = synth_logger(dest_file, csv=True)
     synth_for_words = synth_generator()
     try:
@@ -228,4 +243,5 @@ def generate_audio_for_stories():
     closer()
 
 if __name__ == '__main__':
-    generate_audio_for_stories()
+    generate_audio_for_text_list(['random','education'])
+    # generate_audio_for_stories()