README.md

metadata

title: SOP Audio Analyzer
emoji: 🎙️
colorFrom: blue
colorTo: red
sdk: docker
pinned: false

SOP Audio Analyzer

Test Integrity Analysis - Voice fraud detection for take-at-home tests.

Features

• 🎤 Record or upload audio files
• 🗣️ Speaker diarization - detect multiple voices
• 🎯 Voiceprint extraction - unique ID per speaker
• 🔈 Background analysis - detect whispers, distant voices
• 🤖 Synthetic detection - identify TTS/AI voices
• 📢 Wake word detection - Alexa, Siri, Google
• 🗄️ Cross-test tracking - find same voice across tests

Installation

# Create virtual environment
python -m venv venv
source venv/bin/activate  # Linux/Mac
# or: venv\Scripts\activate  # Windows

# Install dependencies
pip install -r requirements.txt

Run

streamlit run app.py

Project Structure

sop-audio-analyzer/
├── app.py                    # Main Streamlit app
├── requirements.txt
├── src/
│   ├── phase1_foundation/    # VAD, Diarization, Voiceprint
│   ├── phase2_background/    # Background analysis
│   ├── phase6_synthetic/     # Synthetic & wake word detection
│   ├── database/             # SQLite models & queries
│   └── ui/                   # UI components
├── data/
│   ├── db/                   # SQLite database
│   └── clips/                # Extracted audio clips
└── tests/
    └── audio/                # Test audio files

Usage

1. Analyzer tab: Upload or record audio → Analyze → View results
2. Database tab: Browse all voiceprints → Track across tests

Tech Stack

• SpeechBrain: VAD, diarization, speaker recognition
• Whisper: Transcription, wake word detection
• Streamlit: Web UI
• SQLite: Voiceprint database