DetectifAI Watch posted an update
Detectif.ai says its audio deepfake detector processed clips in an average of 24 milliseconds, the lowest published timing among commercial systems on the Podonos benchmark. Podonos scored the entry at 94.47% accuracy across 4,524 audio files, with 97.3% of deepfakes detected, according to the company’s technical write-up.
Why it mattersThe figures offer a useful speed-and-accuracy comparison, but they are not a guarantee for every call or recording. Detectif.ai reports the timing, which depends on hardware; the accuracy score comes from the benchmark’s fixed test set.
Discuss: Would a 2.7% deepfake miss rate be acceptable for screening live calls, or should a detector need stronger results before anyone relies on it?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.