How malevolent-shrine-hq/vocalguard works
VocalGuard is a web console and FastAPI service for detecting synthetic speech and checking speaker identity. Analysts upload recordings or stream microphone audio; the service preprocesses audio, runs deepfake inference, and returns threat assessments, while live sessions also support voiceprint verification and temporal risk scoring. The console also supports authenticated voiceprint enrollment and demo samples. Backend module relationships are supported by the excerpts and SOURCE INDEX; frontend-to-service interaction is documented at a high level in the README, but its specific request wiring was not sampled.
Main entry points: index.py REST and Stream API · streaming_detector.py Streaming Sessions · App.jsx Operations Console · detector.py Audio Preprocessing and Inference
Components 13
Analyst Console
- Operations Console
src/App.jsx - Sign-in UI
src/main.jsx
Detection Service
- REST and Stream API
api/index.py - Session Verification
api/auth.py - Demo Samples
api/index.py
Audio Analysis
- Streaming Sessions
api/streaming_detector.py - Audio Preprocessing and Inference
api/detector.py - Detector Weights
model/voice_deepfake_detector.pth
Identity Verification
- Voiceprint Engine
api/speaker_biometrics.py - Enrolled Voiceprints
model/enrolled_voiceprints.json
Other components
- Analyst
- Clerk Identity Service
- Demo Audio Files
public/samples
Connections 15
- Analyst → Operations Console: uses
README.md - Operations Console → REST and Stream API: submits audio
README.md - Sign-in UI → Operations Console: provides sign-in
src/main.jsx - Sign-in UI → Clerk Identity Service: configures provider
src/main.jsx - REST and Stream API → Session Verification: checks sessions
api/index.py - Session Verification → Clerk Identity Service: verifies tokens
api/auth.py - REST and Stream API → Audio Preprocessing and Inference: runs detection
api/index.py - REST and Stream API → Streaming Sessions: creates sessions
api/index.py - REST and Stream API → Voiceprint Engine: manages profiles
api/index.py - REST and Stream API → Demo Audio Files: serves samples
api/index.py - Streaming Sessions → Audio Preprocessing and Inference: analyzes windows
api/streaming_detector.py - Streaming Sessions → Voiceprint Engine: checks identity
api/streaming_detector.py - Audio Preprocessing and Inference → Detector Weights: loads checkpoint
api/detector.py - Voiceprint Engine → Enrolled Voiceprints: loads and saves
api/speaker_biometrics.py - Demo Samples → Demo Audio Files: reads files
api/index.py