cut-voice-tracks.py allow rewriting transcriptions

scripts/cut-voice-track.py

@@ -55,7 +55,7 @@ def process_chunk(audio_guess, possible_sections):
         return start, end
     
     print('\033[31m' + audio_guess + '\033[0m')
-    print(f'{takes}/u({takes}/*)/d/f/n/h/q')
+    print(f'{takes}/u({takes}/*)/d/f/n/r/h/q')
     while True:
         choice = getch()
         if choice == 'h':
@@ -64,6 +64,7 @@ def process_chunk(audio_guess, possible_sections):
             print(f'u + {takes} - use a take.')
             print(f'u + * -  use all takes as alts.')
             print('f - search ahead for a word or phrase')
+            print('r - rewrite the transcription of this snippet for better matching')
             print('n - repeat a search.')
             print('d - discard this snippet.')
             print('q - save and quit')
@@ -75,6 +76,9 @@ def process_chunk(audio_guess, possible_sections):
         elif choice == 'f':
             cutter.search()
             break
+        elif choice == 'r':
+            cutter.rewrite_transcription(audio_guess, process_chunk)
+            break
         elif choice == 'n':
             cutter.repeat_search()
             break

scripts/util.py

@@ -23,6 +23,8 @@ def args(starting_num, usage, default=None):
 
 class AudioCutter:
     def __init__(self, wav_file, json_file):
+        self.json_file = json_file
+
         # Store a wav file's sound data and json data representing tagged chunks of audio in the wav
         with open(json_file, 'r') as f:
             self.json_info = json.load(f)
@@ -80,6 +82,17 @@ class AudioCutter:
     def repeat_search(self):
         self.searching_for = self.last_search
 
+    def rewrite_transcription(self, audio_tag, chunk_processor=None):
+        info = self.json_info[audio_tag]
+        new_audio_tag = input("new transcription? ")
+        
+        self.json_info[new_audio_tag] = info
+        self.json_info.pop(audio_tag, None)
+        with open(self.json_file, 'w') as f:
+            json.dump(self.json_info, f)
+        if chunk_processor is not None:
+            chunk_processor(new_audio_tag, info)
+
     def process_audio(self, chunk_processor, new_wav_file):
         for (audio_tag, chunk_info) in self.json_info.items():
             # When the AudioCutter is searching for a phrase, skip all audio tags that don't match