Spaces:
Sleeping
Sleeping
Commit History
-
Fix: Apply compatibility patches before speechbrain import
e53bdb6 -
Prioritize root in sys.path and add debug info
35456ff -
Fix Import Error: Add __init__.py files and debug logs
d72a730 -
Fix ModuleNotFoundError by adding root to sys.path
08f0d4c -
Refactor project structure to Unified AI Voice Detection System
b3f89f5 -
Fix SpeechBrain compatibility with newer torchaudio by mocking list_audio_backends
0b7673f -
Fix model loading error by updating torch version and improving logging
6a16e92 -
Heavy & Accurate: Integrated SpeechBrain VAD + MMS-300M pipeline
62f98bb -
Upgrade: Switch to MMS-300M (XLS-R) for robust multilingual deepfake detection
cb14a1d -
Fix: Require heuristic evidence for AI verdict on non-English audio to prevent false positives
0d64f25 -
Implement Conservative Consensus: SNR Analysis and Human Liveness Veto to eliminate false positives
f95b090 -
Implement supreme Hybrid AI detection: sliding window, spectral and temporal features, and language-aware confidence.
04f3da2 -
Add comprehensive debug logging to track AI classification confidence calculations
9d2870b -
Upgrade to specialized deepfake detection model and optimize audio processing
f07936b -
Fix: Align prediction label with pitch variance heuristic
b1df80d -
Optimize: Truncate audio to 5s & add per-step timing logs
28dad85 -
Fix: Final audio processing fixes and enhanced logging
a218549 -
Fix: Add debug logging and robust padding to audio processing
1f10fcd -
Fix: Robust Base64 decoding
4b4453a -
Initial commit: Multilingual Voice Detector with XLS-R Support
2a4d245