Update app.py

app.py

@@ -47,13 +47,6 @@ def transcribe_video(video_path):
 
     return timestamps
 
-def save_transcript_to_json(timestamps, json_file):
-    with open(json_file, 'w') as f:
-        json.dump(timestamps, f, indent=4)
-
-# Initialize the translation pipeline
-translation_pipeline = pipeline("translation", model="Helsinki-NLP/opus-mt-en-{target_language}")
-
 # Function to get the appropriate translation model based on target language
 def get_translation_model(target_language):
     # Map of target languages to their corresponding model names
@@ -65,30 +58,27 @@ def get_translation_model(target_language):
     }
     return model_map.get(target_language, "Helsinki-NLP/opus-mt-en-fr")  # Default to French if not found
 
-def translate_text(timestamps_json, target_language):
+def translate_text(transcription_json, target_language):
     # Load the translation model for the specified target language
     translation_model_id = get_translation_model(target_language)
     logger.debug(f"Translation model: {translation_model_id}")
     translator = pipeline("translation", model=translation_model_id)
 
-    # Parse the input JSON
-    timestamps = json.loads(timestamps_json)
-
     # Prepare output structure
-    translated_timestamps = []
+    translated_json = []
 
     # Translate each sentence and store it with its start time
-    for entry in timestamps:
+    for entry in transcription_json:
         original_text = entry["text"]
         translated_text = translator(original_text)[0]['translation_text']
-        translated_timestamps.append({
+        translated_json.append({
             "start": entry["start"],
             "original": original_text,
             "translated": translated_text
         })
 
     # Return the translated timestamps as a JSON string
-    return json.dumps(translated_timestamps, indent=4)
+    return json.dumps(translated_json, indent=4)
 
 def add_transcript_to_video(video_path, timestamps, output_path):
     # Load the video file
@@ -130,19 +120,15 @@ def upload_and_manage(file, platform, language):
 
     # Define paths for audio and output files
     audio_path = "audio.wav"
-    json_file = "transcript.json"
     output_video_path = "output_video.mp4"
 
     # Transcribe audio from uploaded media file and get timestamps
-    timestamps = transcribe_video(file.name)
-
-    # Save transcript to JSON
-    save_transcript_to_json(timestamps, json_file)
+    transcrption_json = transcribe_video(file.name)
 
-    translated_timestamps = translate_text(timestamps, language)
+    translated_json = translate_text(transcrption_json, language)
     
     # Add transcript to video based on timestamps
-    add_transcript_to_video(file.name, translated_timestamps, output_video_path)
+    add_transcript_to_video(file.name, translated_json, output_video_path)
 
     # Mock posting action (you can implement this as needed)
     post_message = mock_post_to_platform(platform, file.name)
@@ -150,7 +136,7 @@ def upload_and_manage(file, platform, language):
     # Mock analytics generation
     analytics = mock_analytics()
 
-    return post_message, timestamps, json_file, analytics
+    return post_message, transcrption_json, translated_json, analytics
 
 def generate_dashboard(analytics):
     if not analytics:
@@ -177,8 +163,8 @@ def build_interface():
             
             with gr.Row():
                 post_output = gr.Textbox(label="Posting Status", interactive=False)
-                transcription_output = gr.Textbox(label="Transcription Timestamps (JSON)", interactive=False)
-                json_output = gr.Textbox(label="Transcript JSON File", interactive=False)
+                transcription_output = gr.Textbox(label="Transcription JSON File", interactive=False)
+                translated_output = gr.Textbox(label="Translated JSON File", interactive=False)
 
             submit_button.click(upload_and_manage, 
                                 inputs=[file_input, platform_input, language_input],