The effort to catalog the inequity in onscreen roles for women and minorities has a new weapon. The Geena Davis Institute on Gender in Media, at Mount Saint Mary’s University, with financial backing from Google.org, the company’s philanthropic division, will announce on Wednesday a tool that employs video- and audio-recognition technology, along with algorithms, to identify gender, speaking time and additional details about characters presented in films, television shows and other media.
“The research is a tool to help inspire change,” said Madeline Di Nonno, chief executive of the Geena Davis Institute. “It’s not meant to criticize; it’s meant to have the facts so that content creators can be aware and learn from it.”
The software took engineers from the Signal Analysis and Interpretation Laboratory at the University of Southern California and Google two years to develop. Dr. Shri Narayanan, an engineering professor at the university, led the development team there, with input from social scientists and Dr. Caroline Heldman, a professor of politics at Occidental College, along with Hartwig Adam, a Google machine-learning research specialist. “We wanted to see how we can unleash data science, for tools of discovery,” Dr. Narayanan said.
In the first round of research using the tool, a study of the 200 top-grossing, nonanimated films of 2014 and 2015, like “Jurassic World” and “Avengers: Age of Ultron,” found that overall, in 2015, male characters were both seen and heard about twice as much as female characters. Parity on paper does not help: In films with male and female leads, the men nonetheless appear and speak more often than the women. Even in films with female leads, the men still get nearly equal screen and speaking time. Dr. Narayanan will present these findings at symposiums sponsored by the Geena Davis Institute and hosted by Google, in New York and Los Angeles this fall.
“This is such an extraordinary bias-busting tool,” Ms. Crommett said. “A really powerful way to look at the whole picture.”