Update app.py

app.py

@@ -1,39 +1,46 @@
 import gradio as gr
-#from moviepy.editor import VideoFileClip
-from moviepy import VideoFileClip
+from moviepy.editor import VideoFileClip, TextClip, CompositeVideoClip
 from PIL import Image, ImageDraw, ImageFont
 import whisper
 from keybert import KeyBERT
 import numpy as np
+import os
 
-# Load Whisper model and KeyBERT model
+# Load models
 whisper_model = whisper.load_model("base")
 kw_model = KeyBERT()
 
 def process_video(video_path, caption="Your Caption"):
-    # Extract frame at 5 seconds
+    # Load video
     clip = VideoFileClip(video_path)
-    frame = clip.get_frame(5)  # 5 seconds
-    image = Image.fromarray(np.uint8(frame))
 
-    # Add caption
+    # Extract a frame at 5 seconds and save thumbnail
+    frame = clip.get_frame(5)
+    image = Image.fromarray(np.uint8(frame))
     draw = ImageDraw.Draw(image)
     font = ImageFont.load_default()
-    font = ImageFont.load_default()
-  # Make sure Arial.ttf is available
     text_position = (50, image.height - 100)
-    draw.text(text_position, caption, (255, 255, 255), font=font)
-
+    draw.text(text_position, caption, fill="white", font=font)
     thumbnail_path = "thumbnail.jpg"
     image.save(thumbnail_path)
 
-    # Extract keywords
-    result = whisper_model.transcribe(video_path)
+    # Transcribe Hindi audio
+    result = whisper_model.transcribe(video_path, language="hi")
     text = result["text"]
+
+    # Extract keywords from Hindi text
     keywords = kw_model.extract_keywords(text, keyphrase_ngram_range=(1, 2), stop_words='english')
     keywords_list = [kw[0] for kw in keywords]
 
-    return thumbnail_path, ", ".join(keywords_list)
+    # Burn caption into video
+    txt_clip = TextClip(caption, fontsize=50, color='white', font="Arial", bg_color='black').set_position(("center", "bottom")).set_duration(clip.duration)
+    final = CompositeVideoClip([clip, txt_clip])
+
+    # Save processed video
+    output_path = "output_with_caption.mp4"
+    final.write_videofile(output_path, codec='libx264', audio_codec='aac')
+
+    return thumbnail_path, ", ".join(keywords_list), output_path
 
 # Gradio UI
 with gr.Blocks() as demo:
@@ -43,8 +50,12 @@ with gr.Blocks() as demo:
     generate_button = gr.Button("Generate Thumbnail & Keywords")
     thumbnail_output = gr.Image(label="Generated Thumbnail")
     keywords_output = gr.Textbox(label="SEO Keywords")
+    video_output = gr.File(label="Download Processed Video")
 
-    generate_button.click(process_video, inputs=[video_input, caption_input], outputs=[thumbnail_output, keywords_output])
+    generate_button.click(
+        fn=process_video,
+        inputs=[video_input, caption_input],
+        outputs=[thumbnail_output, keywords_output, video_output]
+    )
 
-# Launch in Hugging Face Spaces
 demo.launch()