update

app.py

@@ -3,14 +3,60 @@ import os
 import requests
 import urllib
 from os import path
-from pydub import AudioSegment
+
+import PIL
+from PIL import Image
+from PIL import ImageDraw
+from PIL import ImageFont
+
 
 img_to_text = gr.Blocks.load(name="spaces/pharma/CLIP-Interrogator")
+prompt_to_joke = gr.Blocks.load(name="spaces/ysharma/Voice-to-jokes")
 
+def driver(uploaded_image):
+  #get clip interogator's prompt response
+  prompt = clip_intero_inference(uploaded_image)
+  
+  #get a joke based on the prompt
+  generated_joke = voicetojoke_inference(prompt)
+  
+  #write on the image
+  image0 = write_on_image(uploaded_image, joke)
+  
+  return image0
+  
 def clip_intero_inference(uploaded_image):
   prompt = img_to_text(uploaded_image, fn_index=1)[0]
   return prompt
 
+def voicetojoke_inference(prompt):
+  generated_joke = prompt_to_joke(None, prompt)[1] # dummy_transcript, dummy_audio 
+  return generated_joke 
+
+#def download_image(url):
+#    response = requests.get(url)
+#    return PIL.Image.open(BytesIO(response.content)).convert("RGB").resize((512, 512))
+    
+    
+def write_on_image(uploaded_image, joke): 
+  print("************ Inside write_on_image ***********")
+  #image_path0 = "./distracted0.jpg"
+  image0 = Image.open(uploaded_image).convert("RGB").resize((512, 512)) #(image_path0)
+  I1 = ImageDraw.Draw(image0)
+  myfont = ImageFont.truetype('./font1.ttf', 30)
+  
+  prompt_list = joke #final_solution.split('\n')
+  topline = ' '.join(joke.split()[:5])
+  bottomline = ' '.join(joke.split()[5:])
+  print(f"Joke is : {joke}")
+  print(f"top line is : {topline }")
+  print(f"bottom line is : {bottomline }")
+  
+  I1.text((14, 14), topline, font=myfont, fill =(255, 255, 255))
+  I1.text((14, 454), bottomline, font=myfont, fill =(255, 255, 255))
+  
+  return image0 #, new_prompt
+
 with gr.Blocks() as demo:
   with gr.Row():
     #in_newssource =  gr.Dropdown(["Google News", "The Hindu", "Times Of India"], label='Choose a News Outlet')