Generator traces across languages
A language-disjoint evaluation of forensic representations. Train on 12 languages, then test the same generators on six unseen languages.
91.0%Top-1 generator retrieval on languages excluded from training
Voice model research.
Working preprints on speech forensics and deployment systems. Read the methods, results and limitations of each study.
Language-agnostic detection
Whether generator artifacts survive a change of language, tested with languages removed before training.
Telephone audio
Codecs, restricted bandwidth and noise change the speech signal. Models need evaluation under the conditions in which they will run.
Deployment systems
What detection costs to run on GPU, server CPU and a mid-range Android handset, with detection quality measured on each.
Research studies.
Deepfake detection across devices
One detector, seven deployment configurations. Measure processing capacity and detection quality on GPU, server CPU and Android.
99.65%Of handset test clips processed faster than real time
Evaluate Detectif.ai on your own flow.
Channels, codecs, languages and generators change detection performance.