Update app.py

app.py

@@ -1,15 +1,15 @@
 import streamlit as st
 from TTS.api import TTS
 import torch
-import locale
 import subprocess
+import locale
 import assemblyai as aai
 import requests
-import moviepy.editor as mp
 
 # Load TTS model
 tts = TTS("tts_models/multilingual/multi-dataset/xtts_v2", gpu=True).to("cuda")
 
+
 # Function to translate video
 class Translation:
     def __init__(self, video_path, original_language, target_language):
@@ -42,10 +42,8 @@ class Translation:
             self.tran_code = ''
 
     def extract_audio(self):
-        video = mp.VideoFileClip(self.video_path)
-        audio = video.audio
         audio_path = "output_audio.wav"
-        audio.write_audiofile(audio_path)
+        subprocess.run(['ffmpeg', '-i', self.video_path, '-vn', '-acodec', 'pcm_s16le', '-ar', '44100', '-ac', '2', audio_path])
         st.success("Audio extracted successfully!")
         return audio_path
 
@@ -92,17 +90,8 @@ class Translation:
         translated_text = self.translate_text(transcript_text)
         translated_audio_path = self.generate_audio(translated_text)
 
-        pad_top = 0
-        pad_bottom = 15
-        pad_left = 0
-        pad_right = 0
-        rescaleFactor = 1
-        video_path_fix = f"'{self.video_path}'"
-        audio_path_fix = f"'{translated_audio_path}'"
-        subprocess.run(['python', 'inference.py', '--checkpoint_path', 'checkpoints/wav2lip_gan.pth',
-                        '--face', video_path_fix, '--audio', audio_path_fix, '--pads',
-                        str(pad_top), str(pad_bottom), str(pad_left), str(pad_right),
-                        '--resize_factor', str(rescaleFactor), '--nosmooth', '--outfile', 'output_video.mp4'])
+        # Convert translated audio to video
+        subprocess.run(['ffmpeg', '-i', self.video_path, '-i', translated_audio_path, '-c:v', 'copy', '-map', '0:v:0', '-map', '1:a:0', 'output_video.mp4'])
 
 
 st.title("Translate Your Video")