BoldActionMan commited on
Commit
9558dda
·
verified ·
1 Parent(s): f86b927

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +4 -5
app.py CHANGED
@@ -58,7 +58,6 @@ def process_video(video_file, language_choice):
58
  audio, _ = load_audio(reference_audio, sr=df_state.sr())
59
  enhanced = enhance(model, df_state, audio)
60
  save_audio(reference_audio, enhanced, df_state.sr())
61
- reference_speaker = reference_audio # This is the voice you want to clone
62
 
63
  src_path = os.path.join(output_dir, "tmp.wav")
64
 
@@ -67,7 +66,7 @@ def process_video(video_file, language_choice):
67
 
68
  # Transcribe the original audio with timestamps
69
  sttmodel = whisper.load_model("base")
70
- sttresult = sttmodel.transcribe(reference_speaker, verbose=True)
71
 
72
  # Print the original transcription
73
  print(sttresult["text"])
@@ -116,9 +115,7 @@ def process_video(video_file, language_choice):
116
 
117
  print(f"Final video with subtitles saved to: {final_video_with_subs_path}")
118
  return final_video_with_subs_path, "Video language and language selection are the same, audio not changed."
119
- else:
120
- target_se, audio_name = se_extractor.get_se(reference_speaker, tone_color_converter, vad=False)
121
-
122
  # Choose the target language for translation
123
  language = 'EN_NEWEST'
124
  match language_choice[0:2]:
@@ -178,6 +175,8 @@ def process_video(video_file, language_choice):
178
  batch_segment_files = [item for sublist in batch_segment_files for item in sublist] # Flatten the list
179
 
180
  for segment_file, start, end, translated_text in batch_segment_files:
 
 
181
  # Run the tone color converter
182
  encode_message = "@MyShell"
183
  tone_color_converter.convert(
 
58
  audio, _ = load_audio(reference_audio, sr=df_state.sr())
59
  enhanced = enhance(model, df_state, audio)
60
  save_audio(reference_audio, enhanced, df_state.sr())
 
61
 
62
  src_path = os.path.join(output_dir, "tmp.wav")
63
 
 
66
 
67
  # Transcribe the original audio with timestamps
68
  sttmodel = whisper.load_model("base")
69
+ sttresult = sttmodel.transcribe(reference_audio, verbose=True)
70
 
71
  # Print the original transcription
72
  print(sttresult["text"])
 
115
 
116
  print(f"Final video with subtitles saved to: {final_video_with_subs_path}")
117
  return final_video_with_subs_path, "Video language and language selection are the same, audio not changed."
118
+ else:
 
 
119
  # Choose the target language for translation
120
  language = 'EN_NEWEST'
121
  match language_choice[0:2]:
 
175
  batch_segment_files = [item for sublist in batch_segment_files for item in sublist] # Flatten the list
176
 
177
  for segment_file, start, end, translated_text in batch_segment_files:
178
+ reference_speaker = reference_audio[start:end] # This is the voice you want to clone
179
+ target_se, audio_name = se_extractor.get_se(reference_speaker, tone_color_converter, vad=False)
180
  # Run the tone color converter
181
  encode_message = "@MyShell"
182
  tone_color_converter.convert(