Upload app.py

app.py

@@ -7,28 +7,11 @@ import numpy as np
 import os
 import sys
 from pathlib import Path
+import base64
 
 # Model and Tokenizer Loading
 MODEL_ID = "Qwen/Qwen-Audio-Chat"
 
-# Add the model's directory to sys.path to import its audio module
-def setup_audio_module():
-    try:
-        from huggingface_hub import snapshot_download
-        
-        # Download the model files
-        model_path = snapshot_download(MODEL_ID)
-        if model_path not in sys.path:
-            sys.path.append(model_path)
-            
-        # Now we can import the audio module
-        global Audio
-        from audio import Audio
-        return True
-    except Exception as e:
-        print(f"Error setting up audio module: {e}")
-        return False
-
 def load_model():
     print("Loading model and tokenizer...")
     model = AutoModelForCausalLM.from_pretrained(
@@ -42,15 +25,28 @@ def load_model():
     return model, tokenizer
 
 def process_audio(audio_path):
-    """Process audio file using Qwen's audio module."""
+    """Process audio file for the model."""
     try:
         print(f"Processing audio file: {audio_path}")
-        # Initialize Audio processor
-        audio_processor = Audio()
-        # Process the audio file
-        processed_audio = audio_processor.load_audio_from_file(audio_path)
-        print("Audio processed successfully")
-        return processed_audio
+        # Read audio file
+        audio_data, sample_rate = sf.read(audio_path)
+        
+        # Convert to mono if stereo
+        if len(audio_data.shape) > 1:
+            audio_data = audio_data.mean(axis=1)
+        
+        # Ensure float32 format
+        audio_data = audio_data.astype(np.float32)
+        
+        # Convert to base64
+        audio_bytes = sf.write(file=None, data=audio_data, samplerate=sample_rate, format='WAV')
+        audio_base64 = base64.b64encode(audio_bytes).decode('utf-8')
+        
+        print(f"Audio processed successfully. Sample rate: {sample_rate}, Shape: {audio_data.shape}")
+        return {
+            "audio": audio_base64,
+            "sampling_rate": sample_rate
+        }
     except Exception as e:
         print(f"Error processing audio: {e}")
         return None
@@ -74,13 +70,9 @@ def analyze_audio(audio_path: str, question: str = None) -> str:
     if not os.path.exists(audio_path):
         return f"Audio file not found: {audio_path}"
     
-    # Setup audio module
-    if not setup_audio_module():
-        return "Failed to initialize audio processing module."
-    
     # Process audio
-    processed_audio = process_audio(audio_path)
-    if processed_audio is None:
+    audio_data = process_audio(audio_path)
+    if audio_data is None:
         return "Failed to process the audio file. Please ensure it's a valid audio format."
 
     try:
@@ -94,13 +86,14 @@ def analyze_audio(audio_path: str, question: str = None) -> str:
                 "content": [
                     {
                         "type": "audio",
-                        "audio_data": processed_audio,
+                        "data": audio_data["audio"],
+                        "sampling_rate": audio_data["sampling_rate"]
                     },
                     {
                         "type": "text",
-                        "text": query,
-                    },
-                ],
+                        "text": query
+                    }
+                ]
             }
         ]