This project utilizes a CNN model to differentiate between authentic human speech and AI-generated or cloned audio. It is trained and evaluated using the ASVspoof 2019 LA benchmark dataset.
From the README
--- title: Audio Deepfake Detector emoji: 🎙️ colorFrom: blue colorTo: purple sdk: streamlit sdkversion: "1.58.0" appfile: app.py pinned: false --- 🎙️ Audio Deepfake Detector A CNN-based system that distinguishes genuine human speech from AI-generated / voice-cloned audio, trained and evaluated on the ASVspoof 2019 LA benchmark dataset. Live demo → --- Why this project Voice cloning tools have made synthetic speech nearly indistinguishable from real recordings — a real threat to phone-based authentication, media integrity, and fraud prevention. This project builds and evaluates a spoofing countermeasure: given an audio clip, predict whether it's genuine ("bonafide") or...
Skills
- Python
