.

app.py

@@ -4,6 +4,8 @@ import os
 import torch
 import json
 import time
+import tempfile
+import shutil
 from transformers import AutoTokenizer, AutoModelForCausalLM
 
 # Check if CUDA is available and set the device accordingly
@@ -128,13 +130,30 @@ def classify_and_generate(audio_file):
         if not token:
             return "Error: HF_TOKEN environment variable is not set. Please set your Hugging Face API token."
         
-        # First, classify the audio
-        with open(audio_file, "rb") as f:
-            data = f.read()
+        # Create a temporary file to handle the audio data
+        with tempfile.NamedTemporaryFile(delete=False, suffix='.mp3') as temp_audio:
+            # If audio_file is a tuple (file path and sampling rate)
+            if isinstance(audio_file, tuple):
+                audio_path = audio_file[0]
+            else:
+                audio_path = audio_file
+                
+            # Copy the audio file to our temporary file
+            shutil.copy2(audio_path, temp_audio.name)
+            
+            # Read the temporary file
+            with open(temp_audio.name, "rb") as f:
+                data = f.read()
         
         print("Sending request to Audio Classification API...")
         response = requests.post(AUDIO_API_URL, headers=headers, data=data)
         
+        # Clean up the temporary file
+        try:
+            os.unlink(temp_audio.name)
+        except:
+            pass
+        
         if response.status_code == 200:
             classification_results = response.json()
             # Format classification results