Update app.py

app.py

@@ -4,7 +4,19 @@ lpmc_client = gr.load("seungheondoh/LP-Music-Caps-demo", src="spaces")
 from gradio_client import Client
 
 client = Client("https://ysharma-explore-llamav2-with-tgi.hf.space/")
-sd = gr.load("runwayml/stable-diffusion-v1-5", src="spaces")
+
+from diffusers import DiffusionPipeline
+import torch
+
+pipe = DiffusionPipeline.from_pretrained("stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16, use_safetensors=True, variant="fp16")
+pipe.to("cuda")
+
+# if using torch < 2.0
+# pipe.enable_xformers_memory_efficient_attention()
+
+
+
+
 
 from pydub import AudioSegment
 
@@ -63,9 +75,9 @@ def infer(audio_file):
     
     print(result)
 
-    image = sd(result, fn_index=0)[0]
+    images = pipe(prompt=result).images[0]
     
-    return cap_result, result, image
+    return cap_result, result, images
 
 with gr.Blocks() as demo:
     with gr.Column(elem_id="col-container"):