Update app.py

app.py

@@ -8,11 +8,13 @@ import torch
 pipe1 = pipeline("automatic-speech-recognition", model="openai/whisper-base")
 pipe2 = pipeline("summarization", model="facebook/bart-large-cnn")
 pipe3 = DiffusionPipeline.from_pretrained("stable-diffusion-v1-5/stable-diffusion-v1-5")
+pipe3.to("cuda" if torch.cuda.is_available() else "cpu")
 
 def audio_to_image(audio):
-    transcription = pipe1(audio)["text"]
+    transcription = pipe1(audio)
+    transcription_text = transcription['text']
     
-    summary = pipe2(transcription, max_length=30, min_length=10, do_sample=False)
+    summary = pipe2(transcription_text, max_length=50, min_length=10, do_sample=False)
     summary_text = summary[0]['summary_text']
    
     prompt = summary_text