computing phoneme/word variant for each word in a phrase

generate_similar.py

@@ -124,13 +124,18 @@ def parse_apple_phonemes(ph_str):
     return []
 
 
-def similar_phoneme(ph_str):
+def similar_phoneme_word(ph_str):
     phons = parse_apple_phonemes(ph_str)
     vowels = [i for i in phons if i.vowel]
     random.choice(vowels).adjust_stress()
     return ''.join([str(i) for i in phons])
 
+def similar_phoneme_phrase(ph_str):
+    return ' '.join([similar_phoneme_word(w) for w in ph_str.split()])
 
 def similar_word(word_str):
     similar = pronouncing.rhymes(word_str)
     return random.choice(similar) if len(similar) > 0 else word_str
+
+def similar_phrase(ph_str):
+    return ' '.join([similar_word(w) for w in ph_str.split()])

tts_samplegen.py

@@ -7,14 +7,20 @@ import random
 import os
 import re
 import subprocess
+import time
 import progressbar
 
-from generate_similar import similar_phoneme,similar_word
+from generate_similar import similar_phoneme_phrase,similar_phrase
 
-OUTPUT_NAME = 'rand_edu'
+OUTPUT_NAME = 'go_home'
 dest_dir = os.path.abspath('.') + '/outputs/' + OUTPUT_NAME + '/'
 dest_file = './outputs/' + OUTPUT_NAME + '.csv'
 
+def hms_string(sec_elapsed):
+    h = int(sec_elapsed / (60 * 60))
+    m = int((sec_elapsed % (60 * 60)) / 60)
+    s = sec_elapsed % 60.
+    return "{}:{:>02}:{:>05.2f}".format(h, m, s)
 
 def prog_bar(title):
     widgets = [title, progressbar.Counter(), 'th entry - ', progressbar.FormatLabel(
@@ -104,28 +110,28 @@ class SynthVariant(object):
         orig_phon = self.synth.phonemesFromText_('water')
         return orig_phon != ''
 
-    def generate_audio(self, word, variant):
+    def generate_audio(self, text, variant):
         orig_phon, phoneme, phon_cmd = self.synth.phonemesFromText_(
-            word), '', word
+            text), '', text
         if variant == 'low':
             # self.synth.startSpeakingString_toURL_(word,d_url)
             phoneme = orig_phon
         elif variant == 'medium':
-            phoneme = similar_phoneme(orig_phon)
+            phoneme = similar_phoneme_phrase(orig_phon)
             phon_cmd = '[[inpt PHON]] ' + phoneme
         elif variant == 'high':
-            phoneme = similar_word(word)
+            phoneme = similar_phrase(text)
             phon_cmd = phoneme
         # elif variant == 'long':
         # if phon != '':
         # self.phone_synth.startSpeakingString_toURL_(phon,d_url)
         # else:
         #     self.synth.startSpeakingString_toURL_(word,d_url)
-        fname = dest_filename(word, self.name, self.rate, variant)
+        fname = dest_filename(text, self.name, self.rate, variant)
         d_path, r_path = dest_path(self.name, self.rate, fname)
         # d_url = NSURL.fileURLWithPath_(d_path)
         cli_gen_audio(phon_cmd, self.rate, self.name, d_path)
-        return SynthFile(word, phoneme, r_path, self.name, self.lang, self.rate, variant)
+        return SynthFile(text, phoneme, r_path, self.name, self.lang, self.rate, variant)
 
     def create_synth_dirs(self):
         if self.phoneme_capable:
@@ -170,6 +176,7 @@ def synth_generator():
                 print('Discarding phoneme incapable ', s)
 
     def synth_for_words(words, writer):
+        start_time = time.time()
         prog_title = "Synthesizing {} words : ".format(len(words))
         for s in voice_synths:
             s.create_synth_dirs()
@@ -179,6 +186,9 @@ def synth_generator():
                     update('"{}" with {} variant ({})'.format(w, s, v))
                     synthed = s.generate_audio(w, v)
                     writer(synthed)
+        end_time = time.time()
+        time_str = hms_string(end_time - start_time)
+        print("It took {} to synthsize all variants.".format(time_str))
     return synth_for_words
 
 
@@ -215,9 +225,9 @@ def synth_logger(fname, csv=False):
 
 def generate_audio_for_text_list(text_list):
     (writer, closer) = synth_logger(dest_file, csv=True)
-    synth_for_words = synth_generator()
+    synth_for_texts = synth_generator()
     try:
-        synth_for_words(text_list, writer)
+        synth_for_texts(text_list, writer)
     except:
         import traceback
         import sys
@@ -230,18 +240,19 @@ def generate_audio_for_stories():
     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]
+    text_list = [i for g in stories_data.values() for i in g]
-    (writer, closer) = synth_logger(dest_file, csv=True)
+    generate_audio_for_text_list(text_list)
-    synth_for_words = synth_generator()
+    # (writer, closer) = synth_logger(dest_file, csv=True)
-    try:
+    # synth_for_words = synth_generator()
-        synth_for_words(word_list, writer)
+    # try:
-    except:
+    #     synth_for_words(word_list, writer)
-        import traceback
+    # except:
-        import sys
+    #     import traceback
-        traceback.print_exc(file=sys.stdout)
+    #     import sys
-        pass
+    #     traceback.print_exc(file=sys.stdout)
-    closer()
+    #     pass
+    # closer()
 
 if __name__ == '__main__':
-    generate_audio_for_text_list(['random','education'])
+    generate_audio_for_text_list(['I want to go home','education'])
     # generate_audio_for_stories()