Update app.py

app.py

@@ -6,6 +6,7 @@ import ffmpeg
 import nltk
 import re
 from deep_translator import MyMemoryTranslator
+import num2words
 import soundfile as sf
 from gradio_client import Client, handle_file
 from openvoice_cli.__main__ import tune_one
@@ -245,10 +246,20 @@ def merge_audio_video(video_path, audio_path, output_path="temp_merged.mp4"):
     print(f"✅ Merged video saved to {output_path}")
     return output_path
 
-def main_run(video_path,target_lang):
+def main_run(video_path,target_lang,user_transcript=None, user_translation=None):
     original_audio_file = extract_audio_from_video(video_path)
-    original_text , pace = transcribe_audio(original_audio_file)
-    translated_text = translate_local(original_text,target_lang)
+    if user_transcript:
+        original_text , pace = transcribe_audio(original_audio_file)
+        original_text = user_transcript
+        print(f"Using provided transcript: {original_text}")
+    else:
+        original_text , pace = transcribe_audio(original_audio_file)
+    if user_translation:
+        translated_text = user_translation
+        print(f"Using provided translation: {translated_text}")
+    else:
+        translated_text = translate_local(original_text,target_lang)
+        print(f"Translated Text: {translated_text}")
     translated_audio = synthesize_speech(translated_text, target_lang, "Male", pace)
     synced_translated_audio = match_audio_duration(original_audio_file, translated_audio) 
     cloned_synced_translated_audio = clone_voice(synced_translated_audio, original_audio_file)
@@ -256,12 +267,22 @@ def main_run(video_path,target_lang):
     print(f"✅ Pipeline finished")
     return final_video_nobgm
 
-def audio_pipeline_run(audio_path,target_lang):
-    original_text , pace = transcribe_audio(audio_path)
-    translated_text = translate_local(original_text,target_lang)
+def audio_pipeline_run(audio_path,target_lang,user_transcript=None, user_translation=None):
+    if user_transcript:
+        original_text , pace = transcribe_audio(audio_path)
+        original_text = user_transcript
+        print(f"Using provided transcript: {original_text}")
+    else:
+        original_text , pace = transcribe_audio(audio_path)
+    if user_translation:
+        translated_text = user_translation
+        print(f"Using provided translation: {translated_text}")
+    else:
+        translated_text = translate_local(original_text,target_lang)
+        print(f"Translated Text: {translated_text}")
     translated_audio = synthesize_speech(translated_text, target_lang, "Male", pace)
     synced_translated_audio = match_audio_duration(audio_path, translated_audio) 
-    cloned_synced_translated_audio = clone_voice(synced_translated_audio, original_audio_file)
+    cloned_synced_translated_audio = clone_voice(synced_translated_audio, audio_path)
     print(f"✅ Pipeline finished")
     return cloned_synced_translated_audio
 
@@ -295,30 +316,123 @@ with gr.Blocks(title="Audio/Video Translation Toolkit") as app_interface:
     gr.Markdown("# 🚀 Audio/Video Translation Toolkit")
     with gr.Tabs():
         with gr.Tab("🎬 Translate Video"):
-            with gr.Row():
-                with gr.Column():
+            with gr.Column():
+                with gr.Row():
                     video_in = gr.Video(label="Input Video", height=500)
-                    lang_radio_vid = gr.Radio(choices=["Tamil", "Telugu", "Hindi"], label="Target Language", value="Tamil")
-                    submit_btn_vid = gr.Button("Translate Video", variant="primary")
-                    
-                with gr.Column():
                     video_out = gr.Video(label="Output Video", interactive=False, height=500)
+                with gr.Row():
+                    # Radio buttons for selecting target language
+                    # This allows users to choose one of the mutually exclusive options
+                    lang_radio_vid = gr.Radio(choices=["Tamil", "Telugu", "Hindi"], label="Target Language", value="Tamil")
+                    # Single-select option for mutually exclusive choices
+                    option_select = gr.Radio(choices=["Use my Transcript", "Use my Translation"], label="Optional Input")
+                    # Textboxes for user input, initially hidden
+                user_transcript_vid = gr.Textbox(label="Your English Transcript", lines=3, visible=False)
+                user_translation_vid = gr.Textbox(label="Your Translated Text", lines=3, visible=False)
+                submit_btn_vid = gr.Button("Translate Video", variant="primary")
                     
-            submit_btn_vid.click(fn=main_run, inputs=[video_in, lang_radio_vid], outputs=[video_out])
+                # Toggle visibility based on selected option (only one can be active)
+                option_select.change(
+                    fn=lambda choice: (
+                        gr.update(visible=(choice == "Use my Transcript")),
+                        gr.update(visible=(choice == "Use my Translation")),
+                    ),
+                    inputs=option_select,
+                    outputs=[user_transcript_vid, user_translation_vid],
+                )
+
+            # Include the optional transcript/translation textboxes as inputs (they may be hidden)
+            submit_btn_vid.click(fn=main_run, inputs=[video_in, lang_radio_vid, user_transcript_vid, user_translation_vid], outputs=[video_out])
 
         with gr.Tab("🎵 Translate Audio"):
+            with gr.Column():
+                with gr.Row():
+                    audio_in = gr.Audio(label="Input Audio")
+                    audio_out = gr.Audio(label="Output Audio", interactive=False)
+                with gr.Row():
+                    # Radio buttons for selecting target language
+                    # This allows users to choose one of the mutually exclusive options
+                    lang_radio_aud = gr.Radio(choices=["Tamil", "Telugu", "Hindi"], label="Target Language", value="Tamil")
+                    # Single-select option for mutually exclusive choices
+                    option_select = gr.Radio(choices=["Use my Transcript", "Use my Translation"], label="Optional Input")
+                    # Textboxes for user input, initially hidden
+                user_transcript_aud = gr.Textbox(label="Your English Transcript", lines=3, visible=False)
+                user_translation_aud = gr.Textbox(label="Your Translated Text", lines=3, visible=False)
+                submit_btn_aud = gr.Button("Translate Audio", variant="primary")
+                    
+                # Toggle visibility based on selected option (only one can be active)
+                option_select.change(
+                    fn=lambda choice: (
+                        gr.update(visible=(choice == "Use my Transcript")),
+                        gr.update(visible=(choice == "Use my Translation")),
+                    ),
+                    inputs=option_select,
+                    outputs=[user_transcript_aud, user_translation_aud],
+                )                                
+            submit_btn_aud.click(fn=audio_pipeline_run, inputs=[audio_in, lang_radio_aud, user_transcript_aud, user_translation_aud], outputs=[audio_out])
+            
+        with gr.Tab("✂️ Extract Audio"):
             with gr.Row():
+                video_in_ext = gr.Video(label="Input Video", height=500)
+                audio_out_ext = gr.Audio(label="Extracted Audio")
+            btn_ext = gr.Button("Extract", variant="secondary")
+            btn_ext.click(fn=extract_audio_from_video, inputs=video_in_ext, outputs=audio_out_ext)
+
+        with gr.Tab("✍️ Transcribe"):
+            with gr.Row():
+                audio_in_trans = gr.Audio(type="filepath", label="Input Audio")
                 with gr.Column():
-                    audio_in_pipe = gr.Audio(type="filepath", label="Input Audio")
-                    lang_radio_aud = gr.Radio(choices=["Tamil", "Telugu", "Hindi"], label="Target Language", value="Tamil")
-                    submit_btn_aud = gr.Button("Translate Audio", variant="primary")
-                
+                    text_out_trans = gr.Textbox(label="Transcription")
+                    text_out_pace = gr.Textbox(label="Detected Pace")
+            btn_trans = gr.Button("Transcribe", variant="secondary")
+            btn_trans.click(lambda aud: transcribe_audio(aud), inputs=audio_in_trans, outputs=[text_out_trans, text_out_pace])
+
+        with gr.Tab("🌐 Translate Text"):
+            with gr.Row():
                 with gr.Column():
-                    audio_out_pipe = gr.Audio(label="Output Audio", interactive=False)
-                                
-            submit_btn_aud.click(fn=audio_pipeline_run, inputs=[audio_in_pipe, lang_radio_aud], outputs=[audio_out_pipe])
+                    text_in_tran = gr.Textbox(label="Text to Translate", lines=5)
+                    lang_radio_tran = gr.Radio(choices=["Tamil", "Telugu", "Hindi"], label="Target Language", value="Tamil")
+                    btn_tran = gr.Button("Translate", variant="secondary")
+                text_out_tran = gr.Textbox(label="Translated Text", lines=5, interactive=False)
+            btn_tran.click(fn=translate_local, inputs=[text_in_tran, lang_radio_tran], outputs=text_out_tran)
+
+        with gr.Tab("🔊 Synthesize Speech"):
+            with gr.Column():
+                with gr.Row():
+                    text_in_synth = gr.Textbox(label="Text to Synthesize", lines=5)
+                    audio_out_synth = gr.Audio(label="Synthesized Speech")
+                with gr.Row():
+                    lang_radio_tran = gr.Radio(choices=["Tamil", "Telugu", "Hindi"], label="Target Language", value="Tamil")
+                    gender_radio_tran = gr.Radio(choices=["Male", "Female"], label="Speaker Gender", value="Male")
+                    pace_radio_tran = gr.Radio(choices=["Very_Slow", "Slow", "Normal", "Fast", "Very_Fast"], label="Target Language", value="Normal")
+            btn_synth = gr.Button("Synthesize", variant="secondary")
+            btn_synth.click(fn=synthesize_speech, inputs=[text_in_synth,lang_radio_tran,gender_radio_tran,pace_radio_tran], outputs=audio_out_synth)
+
+        with gr.Tab("⏱️ Sync Duration"):
+            with gr.Row():
+                audio_in_sync1 = gr.Audio(type="filepath", label="Original Audio (for duration reference)")
+                audio_in_sync2 = gr.Audio(type="filepath", label="Translated Audio (to be resized)")
+                audio_out_sync = gr.Audio(label="Duration-Synced Audio")
+            btn_sync = gr.Button("Sync Duration", variant="secondary")
+            btn_sync.click(fn=match_audio_duration, inputs=[audio_in_sync1, audio_in_sync2], outputs=audio_out_sync)
+
+        with gr.Tab("🧬 Clone Voice"):
+            with gr.Row():
+                audio_in_clone1 = gr.Audio(type="filepath", label="Target Audio (e.g., Synthesized Speech)")
+                audio_in_clone2 = gr.Audio(type="filepath", label="Reference Audio (Original Speaker's Voice)")
+                audio_out_clone = gr.Audio(label="Cloned Voice Audio")
+            btn_clone = gr.Button("Clone Voice", variant="secondary")
+            btn_clone.click(fn=clone_voice, inputs=[audio_in_clone1, audio_in_clone2], outputs=audio_out_clone)
+
+        with gr.Tab("🎞️ Replace Audio"):
+            with gr.Row():
+                video_in_rep = gr.Video(label="Input Video", height=500)
+                audio_in_rep = gr.Audio(type="filepath", label="New Audio")
+                video_out_rep = gr.Video(label="Video with Replaced Audio", height=500)
+            btn_rep = gr.Button("Replace Audio", variant="secondary")
+            btn_rep.click(fn=merge_audio_video, inputs=[video_in_rep, audio_in_rep], outputs=video_out_rep)
 
-# --- Launch the App ---
+            # --- Launch the App ---
 if __name__ == "__main__":
     # The launch() method creates a web server and makes the interface accessible.
-    app_interface.launch()
+    app_interface.launch()