Merge branch 'master' of /home/ilml/Public/Repos/speech_scoring

speech_samplegen.py

@@ -12,9 +12,9 @@ import time
 from tqdm import tqdm
 
 from generate_similar import similar_phoneme_phrase,similar_phrase
-from speech_tools import hms_string,create_dir,format_filename
+from speech_tools import hms_string,create_dir,format_filename,reservoir_sample
 
-OUTPUT_NAME = 'story_phrases'
+OUTPUT_NAME = 'test_5_words'
 dest_dir = os.path.abspath('.') + '/outputs/' + OUTPUT_NAME + '/'
 dest_file = './outputs/' + OUTPUT_NAME + '.csv'
 
@@ -227,7 +227,7 @@ def generate_audio_for_stories():
     text_list = sorted(list(set(text_list_dup)))
     generate_audio_for_text_list(text_list)
 
-def generate_test_audio_for_stories():
+def generate_test_audio_for_stories(sample_count=0):
     '''
     Picks a list of words from the wordlist that are not in story words
     and generates the variants
@@ -241,11 +241,12 @@ def generate_test_audio_for_stories():
     word_list = [i.strip('\n_') for i in open('./inputs/wordlist.txt','r').readlines()]
     text_set = set(text_list)
     new_word_list = [i for i in word_list if i not in text_set and len(i) > 4]
-    test_words = new_word_list[:int(len(text_list)/5+1)]
+    # test_words = new_word_list[:int(len(text_list)/5+1)]
+    test_words = reservoir_sample(new_word_list,sample_count) if sample_count > 0 else new_word_list
     generate_audio_for_text_list(test_words)
 
 
 if __name__ == '__main__':
-    # generate_test_audio_for_stories()
+    generate_test_audio_for_stories(5)
     # generate_audio_for_text_list(['I want to go home','education'])
-    generate_audio_for_stories()
+    # generate_audio_for_stories()

speech_segmentgen.py

@@ -57,7 +57,8 @@ class Delegate (NSObject):
         '''Called automatically when the application has launched'''
         print("App Launched!")
         # phrases = story_texts()#random.sample(story_texts(), 100)  #
-        phrases = test_texts(30)
+        # phrases = test_texts(30)
+        phrases = story_words()
         # print(phrases)
         generate_audio(phrases)
 
@@ -174,14 +175,19 @@ class SynthesizerQueue(object):
 
 
 def story_texts():
-    # story_file = './inputs/all_stories_hs.json'
     story_file = './inputs/all_stories.json'
     stories_data = json.load(open(story_file))
-    # text_list_dup = [t[0] for i in stories_data.values() for t in i]
     text_list_dup = [t for i in stories_data.values() for t in i]
     text_list = sorted(list(set(text_list_dup)))
     return text_list
 
+def story_words():
+    story_file = './inputs/all_stories_hs.json'
+    stories_data = json.load(open(story_file))
+    text_list_dup = [t[0] for i in stories_data.values() for t in i]
+    text_list = sorted(list(set(text_list_dup)))
+    return text_list
+
 def test_texts(count=10):
     word_list = [i.strip('\n_') for i in open('./inputs/wordlist.txt','r').readlines()]
     text_list = sorted(random.sample(list(set(word_list)),count))