Tag Archives: deepfake

Jay-Z: not amused.

A mere sample of the potential of deep fake audio – a minor digital polish away from indistinguishable replication. In each case, the voices are entirely computer generated by Tacotron 2, a neural network architecture developed by Google.

Copyright infringement? Jay-Z’s record company thought so.

Follow the Vocal Synthesis Subreddit here.

(Pic: Kevork S. Djansezian/Getty Images)

kottke

Impressionist Jim Meskimen performs his poem ‘Pity the Poor Impressionist’ in the voices of 20 celebrities enhanced by SHAM00K using deepfake imaging.

The video required 1,200 hours of footage, 300,000 images and nearly a terabyte of data to create.

Previously: Jim Meskimen’s Nano Impressions

laughingsquid