Views
No views yet
| Model | Accuracy | F1-Score | AUC-ROC |
|---|---|---|---|
| SVM | 0.92 | 0.92 | 0.95 |
| Logistic Regression | 0.89 | 0.89 | 0.93 |
| Perceptron | 0.85 | 0.85 | 0.88 |
| DNN | 0.94 | 0.94 | 0.96 |
1import pickle
2import librosa
3import numpy as np
4
5# Load model and scaler
6with open('audio_svm_model.pkl', 'rb') as f:
7 model = pickle.load(f)
8with open('audio_scaler.pkl', 'rb') as f:
9 scaler = pickle.load(f)
10
11# Load and process audio
12audio, sr = librosa.load('audio_file.wav', sr=16000)
13features = extract_features(audio, sr) # Use feature extraction function
14features_scaled = scaler.transform(features.reshape(1, -1))
15
16# Predict
17prediction = model.predict(features_scaled)
18probability = model.predict_proba(features_scaled)@dataset{csalt_urdu_deepfake,
title={CSALT Urdu Deepfake Detection Dataset},
author={CSALT},
year={2024}
}