UserGuide.txt

1. Launching the Application:

    After installation, run the app.py file.
    The application will open with a graphical interface.

2. Selecting an Audio File:

    Click the "Select Files" button to open a file dialog.
    Choose an audio file from your system. The supported formats include MP3, WAV, FLAC, AAC, and more.

3. Choosing the Model(s):

    The application provides radio buttons to select one or more models for prediction.
        Random Forest: Traditional machine learning-based detection.
        Melody: Hugging Face transformer model.
        960h: Another Hugging Face model for deeper analysis.
        All: Run all models simultaneously for a more comprehensive analysis.

4. Running the Prediction:

    After selecting the audio file and the model(s), click the "Run Prediction" button.
    The app will begin analyzing the audio, displaying real-time progress in the progress bar.
    Logs will be shown in the text box, and once the analysis is complete, you will see the result with a confidence score.

5. Viewing the Results:

    The result will indicate whether the audio is real or fake.
    The confidence score represents the model’s certainty about the classification.

6. Reviewing Logs and Metadata:

    The application will provide detailed logs about the audio file, such as file format, size, bitrate, and prediction results for each selected model.
    All results and logs are saved for later reference.

7. Visualizing Audio Features:

    Once the analysis is complete, the app will generate visualizations of the audio’s MFCC and Mel Spectrogram, which can provide deeper insight into the audio’s characteristics.