Update app.py

app.py

@@ -1,40 +1,64 @@
-import gradio as gr
+import os
 import librosa
 import numpy as np
 import tensorflow as tf
+from flask import Flask, request, jsonify
 from tensorflow.keras.models import load_model
 
-# Load your pre-trained model (make sure it's in the same directory or provide a path)
-model = load_model('voice_authentication_model.keras')  # Replace with your model's filename
+# Load your pre-trained model (make sure it's in the same directory or provide the correct path)
+MODEL_PATH = 'voice_authentication_model.keras'
+model = load_model(MODEL_PATH)
 
-# Function to extract MFCC features and make a prediction
-def predict_user_or_non_user(audio):
-    # Load the audio file using librosa
-    y, sr = librosa.load(audio, sr=None)  # sr=None to keep the original sampling rate
-    
-    # Extract MFCC features from the audio
-    mfccs = librosa.feature.mfcc(y=y, sr=sr, n_mfcc=13)  # You can adjust n_mfcc as needed
-    mfccs = np.mean(mfccs.T, axis=0)  # Take the mean of MFCCs over time to reduce dimension
-    
-    # Reshape the MFCCs to match the input shape expected by the model
-    mfccs = mfccs.reshape(1, -1)  # Reshape to 1 sample, with the number of features
-    
-    # Predict the class (user or non-user)
-    prediction = model.predict(mfccs)
-    
-    # Convert prediction to readable format
-    if prediction > 0.5:
-        return "User"
-    else:
-        return "Non-User"
+app = Flask(__name__)
 
-# Define the Gradio interface
-iface = gr.Interface(
-    fn=predict_user_or_non_user,  # The function to call when an audio input is given
-    inputs=gr.Audio(type="filepath"),  # Corrected audio input setup
-    outputs="text",  # Output will be text (User or Non-user)
-    live=True  # Live mode so the interface updates in real-time
-)
+# Helper function to extract MFCC features
+def extract_mfcc_features(audio_path):
+    try:
+        # Load the audio file using librosa
+        y, sr = librosa.load(audio_path, sr=None)  # sr=None keeps the original sampling rate
+        
+        # Extract MFCC features
+        mfccs = librosa.feature.mfcc(y=y, sr=sr, n_mfcc=13)  # Adjust n_mfcc as needed
+        mfccs = np.mean(mfccs.T, axis=0)  # Average MFCCs over time to reduce dimensions
+        
+        # Reshape MFCCs for model input
+        return mfccs.reshape(1, -1)  # 1 sample with the number of features
+    except Exception as e:
+        raise ValueError(f"Error processing audio file: {e}")
 
-# Launch the interface
-iface.launch()
+# Define the home route for basic health check
+@app.route('/')
+def home():
+    return "Voice Authentication API is running!"
+
+# Define the prediction endpoint
+@app.route('/predict', methods=['POST'])
+def predict():
+    try:
+        # Check if audio file is present in the request
+        if 'file' not in request.files:
+            return jsonify({"error": "No file provided. Please upload an audio file."}), 400
+        
+        # Save the uploaded file to a temporary location
+        audio_file = request.files['file']
+        file_path = os.path.join("temp_audio.wav")
+        audio_file.save(file_path)
+        
+        # Extract features from the audio file
+        mfccs = extract_mfcc_features(file_path)
+        
+        # Perform prediction
+        prediction = model.predict(mfccs)
+        
+        # Convert prediction to user-readable format
+        result = "User" if prediction > 0.5 else "Non-User"
+        
+        # Clean up the temporary file
+        os.remove(file_path)
+        
+        return jsonify({"prediction": result})
+    except Exception as e:
+        return jsonify({"error": str(e)}), 500
+
+if __name__ == "__main__":
+    app.run(host="0.0.0.0", port=7860)