Forensic Analysis of Synthetic Speech: A Hybrid Approach to Audio Deepfake Detection

Chandrashekhar Naidu, Avani (2025) Forensic Analysis of Synthetic Speech: A Hybrid Approach to Audio Deepfake Detection. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (792kB) | Preview
Preview	PDF (Configuration Manual) Download (1MB) | Preview

Abstract

The spread of artificial speech created by highly sophisticated AI methods posed considerable issues in cybersecurity and forensics. In this work, a hybrid forensic model to detect and analyze deepfake audio is introduced, taking into account systematic differences in acoustic and visual aspects in a speech recording. With the dataset ASVspoof2019 Logical Access, we generate Mel-Frequency Cepstral Coefficients (MFCCs) and mel-spectrogram to create a Random Forest (RF) classifier and a Convolutional Neural Network (CNN), respectively. The RF model gives an interpretable answer in terms of SHAP value analysis, whereas the CNN is used through Grad-CAM to find localizable discriminative spectral characteristics pointing to the presence of spoofed speech. In order to make detection more robust, we add more measures of voice quality beyond PESQ, e.g., jitter, shimmer, pitch variability, and harmonics-to-noise ratio that are routinely used in speech pathology, but are not widespread in spoofing detection pipelines. Our proposed hybrid strategy shows an improved rate of accuracy and interpretability when it comes to differentiating authentic and spoofed sound. The presented framework runs through a Streamlit browser-based interface, which allows real- time audio signals, model decision visualization, and clear forensic investigation. This study is of value in the emerging discipline of explainable AI in digital forensics, that paired with a reasonable and explainable solution is effective in mitigating audio-based cyber threats.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Mustafa, Raza Ul UNSPECIFIED
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science Q Science > QH Natural history > QH301 Biology > Methods of research. Technique. Experimental biology > Data processing. Bioinformatics > Artificial intelligence Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Artificial intelligence Q Science > QA Mathematics > Computer software > Computer Security T Technology > T Technology (General) > Information Technology > Computer software > Computer Security
Divisions:	School of Computing > Master of Science in Cyber Security
Depositing User:	Ciara O'Brien
Date Deposited:	17 Aug 2026 14:03
Last Modified:	17 Aug 2026 14:03
URI:	https://norma.ncirl.ie/id/eprint/9526

Actions (login required)

View Item