Update app.py

app.py

@@ -430,35 +430,69 @@ def add_transcript_voiceover(video_path, translated_json, output_path, add_voice
 
     logger.info("Video processing completed successfully.")
 
-# Voice cloning function with debug and error handling
+import os
+import traceback
+from TTS.api import TTS
+import wave
+import logging
+
+logger = logging.getLogger(__name__)
+
+# Initialize TTS model only once (outside the function)
+tts = TTS(model_name="tts_models/multilingual/multi-dataset/xtts_v2")
+
 def generate_voiceover_clone(translated_json, desired_duration, target_language, speaker_wav_path, output_audio_path):
     try:
-        full_text = " ".join(entry["translated"] for entry in translated_json)
-        speed_tts = calculate_speed(full_text, desired_duration)
+        # 1. Assemble full text
+        full_text = " ".join(entry["translated"] for entry in translated_json if "translated" in entry and entry["translated"].strip())
+
+        if not full_text.strip():
+            logger.error("❌ Translated text is empty. Skipping TTS generation.")
+            return None, "❌ Translated text is empty."
+
+        # 2. Check speaker file path
         if not speaker_wav_path or not os.path.exists(speaker_wav_path):
-            return None, "❌ Please upload a valid speaker audio file."
+            logger.error(f"❌ Speaker WAV path not found: {speaker_wav_path}")
+            return None, f"❌ Speaker audio not found: {speaker_wav_path}"
+
+        # Optional: Print speaker audio duration
+        try:
+            with wave.open(speaker_wav_path, 'rb') as wav_file:
+                duration = wav_file.getnframes() / wav_file.getframerate()
+                logger.info(f"🔊 Speaker WAV Duration: {duration:.2f}s")
+        except Exception as e:
+            logger.warning(f"⚠️ Could not read speaker WAV duration: {e}")
 
-        print(f"📥 Received text: {full_text}")
-        print(f"📁 Speaker audio path: {speaker_wav_path}")
-        print(f"🌐 Selected language: {target_language}")
-        print(f"⏱️ Target speed: {speed_tts}")
+        # 3. Log key inputs
+        logger.info(f"📥 Received Text: {full_text}")
+        logger.info(f"📁 Speaker WAV Path: {speaker_wav_path}")
+        logger.info(f"🌐 Target Language: {target_language}")
+        logger.info(f"💾 Output Path: {output_audio_path}")
+        logger.info(f"⏱️ Target Duration: {desired_duration:.2f}s")
 
-        # Run TTS with speed control (if supported by model)
+        # 4. Call TTS to generate audio
         tts.tts_to_file(
             text=full_text,
             speaker_wav=speaker_wav_path,
-            language=language,
-            file_path=output_audio_path,
-            speed=speed_tts  # <- add speed control
+            language=target_language,
+            file_path=output_audio_path
+            # Uncomment if your model supports speed:
+            # speed=speed_tts
         )
-        print("✅ Voice cloning completed.")
+
+        # 5. Confirm file was written
+        if not os.path.exists(output_audio_path):
+            logger.error(f"❌ File NOT generated after tts_to_file: {output_audio_path}")
+            return None, f"❌ Voiceover file not generated at: {output_audio_path}"
+
+        logger.info("✅ Voice cloning completed successfully.")
         return output_audio_path, "✅ Voice cloning completed successfully."
 
     except Exception as e:
-        print("❌ Error during voice cloning:")
-        traceback.print_exc()
-        error_msg = f"❌ An error occurred: {str(e)}"
-        return None, error_msg
+        logger.error("❌ Error during voice cloning:")
+        logger.error(traceback.format_exc())
+        return None, f"❌ An error occurred: {str(e)}"
+
 
 def truncated_linear(x):
     if x < 15: