Shodh AI is a unified security dashboard that detects digital manipulation in Video, Audio, and Images, and verifies spoken content against global news sources.
- Visual Deepfake Detection — EfficientNetV2-S model (fine-tuned on FaceForensics++) identifies face manipulation in images and videos.
- Audio Authenticity Scan — Detects synthetic/AI-generated voices using Audio Spectrogram Transformers.
- Speech-to-Text Transcription — Powered by OpenAI Whisper for high-accuracy script extraction.
- Editable Transcript Verification — Review and correct the AI transcript before fact-checking.
- News Fact-Checker — Cross-references transcripts with Google News RSS to calculate a truth/correlation score.
- Unified Interface — One upload. One click. Full analysis.
git clone https://github.com/YOUR_USERNAME/Misinformation-detector.git
cd Misinformation-detectorpython -m venv .venv
# Windows
.venv\Scripts\activate
# macOS/Linux
source .venv/bin/activatepip install -r requirements.txtstreamlit run shodh_dashboard.pyThis project is ready for one-click deployment on Streamlit Cloud:
- Push your repo to GitHub.
- Go to share.streamlit.io and sign in with GitHub.
- Click New app → Select your repo → Set Main file to
shodh_dashboard.py. - Click Deploy.
The
packages.txtfile automatically installs system dependencies (ffmpeg, libsndfile) on the cloud server.
| File | Description |
|---|---|
shodh_dashboard.py |
Main Streamlit application (cloud-ready) |
scanner_app.py |
Desktop auto-scanner (PyQt6, local use only) |
weight.pth |
Trained deepfake detection weights (~80 MB) |
blaze_face_short_range.tflite |
MediaPipe face detection model |
requirements.txt |
Python dependencies |
packages.txt |
System-level apt dependencies for Streamlit Cloud |
.streamlit/config.toml |
Streamlit theme and server configuration |
This tool is intended for research and educational purposes. Always verify AI results with official sources.