Blog

Technical write-ups on audio deepfake detection, benchmarks and deployment.

  1. BenchmarkHow We Became S.O.T.A. on the Podonos Independent Deepfake Detection BenchmarkA technical write-up of the Podonos audio deepfake benchmark: 24 ms per clip, the lowest reported commercial latency, with 94.47% accuracy across all 4,524 files.5 min read