Preprint
Aug 2026
The Last Mile of Deepfake Speech Detection: An Industry-Academia Experience Report
Based on a three-year effort with Phonexia, a commercial speaker-recognition vendor, barriers encountered while building and deploying a detector are connected to concrete research and coordination proposals: shared standards for commercially usable datasets, realistic deployment benchmarks, and scores that non- expert...
Anton Firc, Kamil Malinka, V. Stanek et al.
· 0 citations