Update app.py

app.py

@@ -5,6 +5,7 @@ from gtts import gTTS
 from pydub import AudioSegment
 from transformers import GPT2LMHeadModel, GPT2Tokenizer, pipeline
 from PIL import Image
+import random
 
 generator = pipeline('text-generation', model="checkpoints")
 tokenizer = GPT2Tokenizer.from_pretrained('checkpoints')
@@ -31,7 +32,7 @@ def generate_output(image, name, date_of_birth):
     except Exception as e:
         return None, f"No se pudo generar el audio: {str(e)}"
 
-    command = f"python3 inference.py --checkpoint_path checkpoints/wav2lip_gan.pth --face {image} --audio audio.wav --outfile video.mp4 --nosmooth"
+    command = f"python3 inference.py --checkpoint_path checkpoints/wav2lip_gan.pth --face {image.name} --audio audio.wav --outfile video.mp4 --nosmooth"
     process = subprocess.run(command, shell=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
     if process.returncode != 0:
         error_message = process.stderr.decode("utf-8")
@@ -50,16 +51,10 @@ def error_message_fn(error_message):
     else:
         return None
 
-def get_image_options():
-    image_options = []
-    for i in range(1, 7):
-        image_options.append({"value": f"{i}.jpg", "label": f"Imagen {i}"})
-    return image_options
-
 iface = gr.Interface(
     fn=generate_output,
     inputs=[
-        gr.inputs.Dropdown(label="Selecciona una imagen:", choices=get_image_options()),
+        gr.inputs.Image(label="Imagen", type="file"),
         gr.inputs.Textbox(lines=1, label="Nombre", placeholder="Ingresa tu nombre"),
         gr.inputs.Textbox(lines=1, label="Fecha de Nacimiento", placeholder="DD/MM/AAAA")
     ],