This will make you wink

Radiolab did a fascinating 50 minute episode about new technology. Adobe Corp., known for PDF files and Photoshop, wanted a Photoshop for sound, a tool to make movie sound editing, especially speech editing, easier. They developed a program to break an actor’s speech down into its component sounds. Then using a text editor they could edit what the actor said and the program would create the actor’s voice saying the new words. They were able to insert words into the text that the actor had not spoken, as long as the actor had spoken all the component sounds.

Another development was by a team at the University of Southern California. They could focus the camera on a person and using that as a guide in real time manipulate the facial features of an image of a person. Whatever the human did – grimace, raise eyebrows, whatever – the image copied. It is like controlling a puppet. This allows a company to make a commercial featuring a famous actor and send it overseas. In the new country, perhaps China, this program would allow moving the lips so it looked like the famous actor was speaking Mandarin. Couple the lips with the first program and the Mandarin would be in the actor’s voice.

Pretty cool!

Now consider the voice that was broken down and recreated and the image being manipulated belonged to President Obama. And the people writing the text and pulling the puppet strings are purveyors of fake news. It now becomes easy for the perpetrators to produce a video showing Obama saying all kinds of nasty things. Those who study manipulated images would probably be able to identify that the video was fake.

But the common man would not, and probably wouldn’t hear about the fakery until after the video had done its damage.

So what is true? We are leaving a difficult problem for the next generation to solve.