English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

The task of deepfakes detection is far from being solved by speech or visionresearchers. Several publicly available databases of fake synthetic video andspeech were built to aid the development of detection methods. However,existing databases typically focus on visual or voice modalities and provide noproof that their deepfakes can in fact impersonate any real person. In thispaper, we present the first realistic audio-visual database of deepfakesSWAN-DF, where lips and speech are well synchronized and video have high visualand audio qualities. We took the publicly available SWAN dataset of real videoswith different identities to create audio-visual deepfakes using several modelsfrom DeepFaceLab and blending techniques for face swapping and HiFiVC, DiffVC,YourTTS, and FreeVC models for voice conversion. From the publicly availablespeech dataset LibriTTS, we also created a separate database of only audiodeepfakes LibriTTS-DF using several latest text to speech methods: YourTTS,Adaspeech, and TorToiSe. We demonstrate the vulnerability of a state of the artspeaker recognition system, such as ECAPA-TDNN-based model from SpeechBrain, tothe synthetic voices. Similarly, we tested face recognition system based on theMobileFaceNet architecture to several variants of our visual deepfakes. Thevulnerability assessment show that by tuning the existing pretrained deepfakemodels to specific identities, one can successfully spoof the face and speakerrecognition systems in more than 90% of the time and achieve a very realisticlooking and sounding fake video of a given person.

Vulnerability of Automatic Identity Recognition to Audio-Visual Deepfakes