A 500 million parameter deepfake audio detection model from the Speech Arena 2025 benchmark.
No lifecycle dates recorded.
No resource links recorded.